Jarvis MLis in the AI Infrastructure Industry

Founded by the former Google MLops team behind Google Ads and Payments, Aidaptive is powering the next generation of digital commerce with an end-to-end autonomous intelligence platform.

Jarvis ML

jarvisml

An Open Essay on the U.S. Federal Reserve, Treasuries, Cryptocurrencies and More: Part 3

When Jerome Powell speaks, traders get weak in the knees. His presence paralyzes the thought process of the most seasoned veterans. Everywhere I turn it’s Jerome this, Powell that. It’s clear that market gossip is the hot take of the day. What was once a discussion about tether flows or exchange flows is now a discussion about inflation, yields, and the amount of rate hikes Wall Street expects. Meanwhile, JPow has yet to even start raising rates. The market has been acting this way for over a month now, and quite honestly it feels like we have overdone it. To decide if we are in fact overdoing it, I decided to do a three part series on the macro economy. In this series we look at Eurodollars, treasuries, inflation, and the federal funds rate through time. The goal is simple… Is this time really that much different? Are there similarities to borrow from history? Can we create a framework to help guide us through the JPow gossip columns of Twitter? Answering these questions is how we gain a leg up in the market. And today in Part-two we look at Treasuries through time to answer these questions. For those that missed Part-One: I highly recommend you start there. While not essential reading, I will be borrowing the template we laid out in that analysis to dissect the treasury market today. The Market’s Next Phase Then after today we should have a nice framework in place for how markets react in the face of rate hikes. Then in Part Three we can see if the framework breaks down… Because that’s where I think everything starts to get pear shaped. OK, so before we begin let’s quickly refresh ourselves on what the template looked like from Part One: 1. Futures expanding and rates at zero → Steadily bullish off lows / Easy Street 2. Futures contracting and rates rising off zero → choppy / semi-bullish market 3. Futures expanding with plateauing rates → very bullish / meltup incoming 4. Futures expanding with rates dropping → it’s over Now the main difference this time is we will substitute the word futures with treasuries. After all, we are looking at the yield on U.S. Government treasuries. These treasuries have a payback term from as little as one month to thirty years, and in the charts below we plot these yields all on the same chart. This stacked representation of yields lets us see them respond as a group over time to changes in the Federal Funds Rate. This Federal Funds Rate is what changes when the market discusses rate hikes by JPow on Company. And for our purposes. we will keep this analysis fairly rudimentary. No talk of flattening and steepening, nor the bearish and bullish varieties. Instead, let’s view how these yields contract and expand in relation to one another just like we saw with Eurodollar futures. And in this way we can compare it to the Federal Funds Rate and the Nasdaq to see if there are any patterns here. Let’s get started… First up, let’s look back to the post dot-com bubble. In the wake of the dot-com bubble rates dropped to nearly zero in short order. The response in the treasury market is shown below with yields expanding. That can be seen on the left-hand side of the chart below. When I say expand I mean the colorful lines are spread apart. When rates were low, yields fully expanded. And the market went on to march higher with low volatility. Using our template from earlier this is known as this “Restart” or Easy Street. 1. Treasuries expanding and rates at zero → Steadily bullish off lows / Easy Street In 2004, the market began to expect higher rates. As the market started to expect higher rates, the Nasdaq got soft. That’s the first vertical white line above. This is a situation much like we see today. But once we finally saw rates actually rise… Meaning JPow is no longer just talking the talk, but walking the walk… yields transitioned from an expansion period to contracting. This is labeled as “Lift off” in the chart above - rates began to rise. This period of rates rising and yields compressing was not great for passive Nasdaq holders. It was a stock pickers market as well. And indicated a preference for realizing profits when earned. To use our framework: 2. Treasuries contracting and rates rising off zero → choppy / semi-bullish market Once yields were fully compressed and actually started to cross over one another, rates stopped hiking. The market responded to this pause in rate hikes with bullishness. The market ripped higher in the period labeled as “The Meltup”.  When the end of the meltup phase was nearing an end we saw: 3. Treasuries expanding with plateauing rates → very bullish / meltup incoming The final stage was when the Federal Reserve dropped rates, yields fully expanded, and the floor came out of the market. 4. Treasuries expanding with rates dropping → it’s over This represented a full cycle from low rates to raising rates and back down to zero rates. And in short order the cycle reset and we were back at it… Here now is the post-2008 time period. In 2009 yields fully expanded. Initially returns were simple during the “Restart” phase. Moving into “Lift Off” yields began to compress with the Nasdaq turning soft just prior. Soon after the market returned to its semi-bullish trajectory, but with higher volatility. Once yields fully compressed the meltup began as rates stopped rising. This is when the yields started to expand again. Then the final stage ended with rates dropping and yields expanding again. You can view this pattern in the chart below. Again, we say this type of sequence with Eurodollars as well in . When the futures were fully expanded, returns were better. And as things contracted, there was more volatility. And once the curves begin to overlap and look messy, things went haywire. Part One’s issue Moving the analysis with an eye towards today, we can see where we are in the cycle. We are at “Lift Off”. Yields are starting to anticipate higher rates in the coming month. This results in yields contracting and volatility being higher. I don’t think you need me to tell you that volatility is up since this lift off began. It’s important to know that looking back to prior instances, this softness can take place with some growth ahead. Assuming the initial overreaction is now out of the way, the market will dust itself off and reallocate capital to Nasdaq / crypto opportunities. But here’s the thing… Notice how the yellow line at the top acts as a resistance for the rest of the yields. Treasuries with maturities less than 30-year periods were never really able to push beyond the yield the market gave for 30-year maturities. Intuitively this makes sense… Why lend out capital at 3% for 30-years when you can do it for 2-year? What is the incentive to tie myself to the longer maturity? That’s why the 30-year yield line acts as resistance. But here’s the thing… 30-year yields are trending lower and lower. They are making lower highs historically. Here’s a chart showing 30-year yields since the 1980’s in comparison to the Federal Funds Rate. This is very important in evaluating how long this next cycle will last. Because historically the Federal Reserve needs the 30-year yields to rise to avoid “things breaking”. That’s because of what we saw earlier. When yields compress, “It’s Over” phase quickly unfolds. And based upon the pattern in the chart above as rates exceed 2.5%, things are in a new territory. And it likely won't last which is why 30-year yields are showing atrophy. Meaning rates on treasuries are deteriorating overtime. They lack strength. Translating this into how it impacts the Federal Reserve, they will likely stop hiking rates in short order. They simply don’t have much room before yields contract to the point where things get wonky... Which is actually very close to where we already find ourselves today. This is the main issue we need to grapple with in the coming cycle. And it’s something that will be a subject for Part Three. In the meantime, don’t let the JPow cloud your judgement. Let him raise rates. And instead focus on where to allocate your next slug. Your Pulse on Crypto, Ben Lilly Also Published here P.S. - If you want to tune in to our weekly market update on crypto, head over to and subscribe. This way you can read the most timely takes as they go out. Also, be sure to check us out at to learn more about our services and team. https://jarvislabs.substack.com/ www.jarvis-labs.xyz

An Open Essay on the U.S. Federal Reserve, Treasuries, Cryptocurrencies and More: Part 2

As I write these words, I have no idea where the analysis is about to take me. I’ve loaded my Tradingview software with about 30 various eurodollar futures contracts, about a dozen US treasury yields, the federal funds rate, inflation, and equity indexes. The goal in doing so is to create a frame of mind for how the next year or two looks like through the lens of two primary things: the U.S. Federal Reserve and debt. From there I believe we can start making some assumptions about bitcoin and cryptocurrencies as a whole. Then track those assumptions as they begin to play out or prove to be wrong. And if we lay it all out here we now have a reference point to understand how things are unfolding as it happens. The motivation for this came after Arthur Hayes wrote his piece called, “ ”. I got a flurry of messages from people thinking the world was ending. Instead of giving my opinion I decided to do some research of my own. Maelstrom And based upon my own interpretation of the market before I start this essay, the word maelstrom seems spot on. It means a state of confused movements. The FED has transitioned in the first half of 2022 from a stance that helps inflate asset prices to one that seems like it might cause wreckage, investors across the board all have seeds of doubt - including me. So I wanted to do what I enjoy most, writing it down to form my own conclusion. To get started I decided the macro indicators listed earlier are a good first stop for research. My hunch before I get too far along here is that this is going to be a three part essay. It will be too much to go at it all at once. Which is why breaking this up into sections might help. It also might help me in using enough ink to explain things in an easier manner. For today’s first essay the topic I’ll focus on is Eurodollars. I know, it’s making you confused already, but we will go slow. Next, we can move into Treasuries, and then end it with a discussion on how unique today's setup is once we loop inflation into the mix. Now, one thing to keep in mind before we start is my base assumption throughout my analysis is that the world is U.S. centric. While central banks around the world might look to front-run policies being discussed by the U.S. Federal Reserve (FED), these decisions are being made with the FED in mind. Even China makes currency conversions with the USD in mind. Additionally, with the world being U.S. centric that means cryptocurrencies are reactionary to a lot of the macro policies employed by the FED. With that out of the way, let’s figure out what the heck is going on today and what the next couple of years might look like by looking back through history. A Helping of Veggies Eurodollars… One of the most opaque and yet insanely massive part of the financial system. It’s the less regulated version of the U.S. dollar that resides in Europe. Explaining the way these U.S. dollar aboard are birthed would take up a lot of time - and frankly, I don’t even fully understand the accounting methods that give rise to their existence. But for perspective in how important they are, the December 2022 Eurodollar contracts have 1.5 million contracts of open interest. Each contract is $1 million in size. Which means for one contract we are talking trillions of dollars of interest. I don’t think I need to tell you that’s a lot. In terms of why these contracts are even traded in the first place, I’ll use Chicago Mercantile Group’s explanation, “Eurodollar futures and options are the preferred tool of traders to express views on future interest rate moves”. Summed up, Eurodollar futures are an expectation for what interest rates will be at the time of their expiration. If you think rates in some year down the line will go up or down for whatever reason, then you can use these contracts to profit. Now, the interest rate in question that Eurodollars track is actually the LIBOR rate… yawn. All you need to know is LIBOR is another type of borrowing rate created between banks. And this rate tends to be about 10-15bps greater than the Fed Funds rate. But to keep the essay simple and convenient for myself I will refer to Eurodollar futures as the FED’s expected rate at the time of expiry. Said another way, for our purposes here I will use Eurodollar futures as a proxy for future expectation of FED rates. For a quick example on how we can interpret what the price of contract means, right now the December 2022 contract is 96.195. If we take 100 and subtract 96.195 from it, we get 3.805. Which tells us in December the market thinks rates will be at 3.8%, higher than where we are today. Why does that matter? The FED has been aggressive about raising rate for a couple months now. So the question comes up whether or not the market has priced in these expected hikes or not. Eurodollars… Exhilarating, am I right? I know it’s probably one of the most boring things I’ve ever talked about. But we need to eat our veggies here if we want to be strong enough to endure the crypto shock waves of 2022 and beyond. And in terms of crypto, when these rate hikes become less aggressive, expect capital to get deployed into the industry. The key is for the expectations to match up with what unfolds. And as we will see shortly, there is a good reason to time up this event because the market may give us a very bullish event soon. The Rhyme Let’s pull up the current Eurodollar’s chart again. Notice how the various lines are converging. What does this mean? It is an indication the market really has no idea what to expect after the end of 2023. That’s because 2023-2030 expiration dates are all bunched up around the same price, forming a steaming pile of indecision. Maelstrom again… And whatever direction the contracts move next - after they they decide they don’t need to follow the heard - will really dictate what happens next. In terms of what happens next, this could mean several things. They could all consolidate, they could contract, they could trend up, or even down. To figure out how to interpret how direction matters let’s look back at at a time when rates were near zero, Eurodollars were converging (like above), and the FED was increasing rates. It’s a situation similar to what we are facing today. The dot-com meltdown. Take a second and check out the chart below. I know all of these lines are a lot to take in. I don’t expect you to understand the nuances of them just yet. What I want you to do is simply focus on when these multi-colored lines expand and contract. That’s it. Just like markets expand and contract, so do Eurodollar futures curves. Though, in the curve sense it’s more along the lines of flattening/steepening. But again, no need to get caught up in this as I’m sure this is already making your head spin enough. Contract / Expand. That’s our focus. In the chart below, notice how the multi-color lines around 2002 get spread out as the white line at the bottom (labeled as Fed Fund rates) approaches zero… I think of this as the start of a new cycle. A time period after stocks crashed and the FED is now returning back to zero rates. A reset. This transition back to the floor where interest rates are near zero and market prices are low is the FED restarting the economy after a blow. So again, rates hit the floor… while the multi-color lines that represent Eurodollars futures contracts expand. Rates at zero, Eurodollars expanding… Meanwhile, the white line at the top - the Nasdaq index - has great returns with low volatility. It’s a time I like to call ‘Easy Street’. It’s a steady return period causing low levels of emotional distress period. Now, in order to summarize it for a template, let’s call this time period stage one or the restart: 1. Futures expanding and rates at zero → Steadily bullish off lows / Easy Street Here’s the chart showing it. Now, on a quick side note… I’m using the Nasdaq in this chart because nearly two decades ago we could say this was Web1 and Web2 stocks. And if crypto’s narrative today is Web3, then this should act as a decent proxy for this analysis without getting too distracted with alternative crypto narratives like store-of-value, hedge to U.S. monetary policy, or fixed supply. Ok, with that out of the way let’s look at when rates started to rise in 2004. A time period I’m calling the second phase. At this time Eurodollars began to consolidate, while the Nasdaq got choppy. Let’s classify this second phase: 2. Futures contracting and rates rising off zero → choppy / semi-bullish market Here is how it looks in a chart: The FED continued to hike until the middle of 2006 when the index sold off. This selloff - which we can attribute partially to the recession warning of 10Yr treasury yields dropping below 2Yr yields - was the market beginning to buckle under these higher rates. In response, monetary policy soften with the FED ending their rate hike regime. This is seen in the white Federal Funds Rate line at the bottom of the chart going sideways for 2007, and starting to drop again. Meanwhile, Eurodollars expand again. Contract, expand, contract, expand - see a pattern? Anyways, here’s the chart of what I’m calling stage three - the meltup. 3. Futures expanding with plateauing rates → very bullish / meltup incoming The expanding Eurodollars were the sign that the financial system was starting to expect lower rates. And soon after the FED began to deliver. Which at first, a rate hike is met with investors backing up the truck. It’s two servings at the buffet table euphoria. But the damage was already done to the financial system. The FED saw the warning signs and continued to drop rates even while the market ripped. The part was over. 4. Futures expanding with rates dropping → it’s over OK, so we went through an entire cycle with rate back at the floor and Eurodollars expanding again - “1”. So to quickly reiterate what just happened, here’s our initial template. The takeaways. 1. Futures expanding and rates at zero → Steadily bullish off lows / Easy Street 2. Futures contracting and rates rising off zero → choppy / semi-bullish market 3. Futures expanding with plateauing rates → very bullish / meltup incoming 4. Futures expanding with rates dropping → it’s over Let’s see if this checks out during the post-2008 market debacle. The Second Verse Let’s move a bit faster through this next time period. A time when the financial system was maxed out on debt and leverage. Below the timeline starts where we left off in the last chart - 2008 crash. What we see in 2009 once rates hit zero was the last part of the small cycle… 1. Futures expanding and rates at zero → Steadily bullish off lows / Easy Street Then phase two began in mid-2010 as a very long Eurodollars consolidation period got underway. 2. Futures contracting and rates rising off zero → choppy / semi-bullish market What’s important to note here is that during “Phase Two” there was a contraction happening on part of the Eurodollars… But not any movement at the top. This was a false signal. If you look back at Phase Two in the previous cycle you’ll notice the lines at the top trended down as the bottom lines did. This is why Phase Two here is actually seven years in length. Rates went sideways at the end of 2018 when trouble showed up… Eurodollars began to expand a bit - setting us up for “The Meltup” stage or “3”. And then finally the FED capitulation and dropped rates to zero with Eurodollars expanding again with the nearest expirations topping out near zero rates. To sum up what we see in the chart below… 2. Futures contracting and rates rising off zero → choppy / semi-bullish market 3. Futures expanding with plateauing rates → very bullish / meltup incoming 4. Futures expanding with rates dropping → it’s over With the cycle ending in early 2020 at the start of the pandemic, we entered the first stage of the cycle again. This means we are likely to see rates rising again… Which are already underway. But the reality is, how far can the FED go before it needs to end “Phase Two” and stop hiking rates? Looking at rates over time we see the FED is not able to raise it past previous highs. It’s a slow trend towards zero. Based upon this the FED might need to step back in after it goes above 1.75%. Going back to our Eurodollars framework again, and applying this rate expectation, here’s where we are… 2. Futures contracting and rates rising off zero → choppy / semi-bullish market Based on history, the market still has runway for higher highs. And applying this Nasdaq proxy to crypto, it alludes to higher volatility with mediocre growth… And spits of strong growth. These spits of strong growth in Phase Two will likely happen as false signals show. Regardless, this environment is better suited to a trader and stock picker than the multi-year hodler. Then once we see a Nasdaq selloff and the FED plateauing rates again… Get ready for a ride towards the meltup. It’s a bubble that surely be an absolute monster. On a final note… I hope this essay serves as clarity. One that helps you poke holes in analysts claiming the “end is near”, which mostly stems from arguments centering on the FED raising rates. Sure, we’ll see drawdowns, but the macro cycle has runway if this framework continues to repeat like it has over the last twenty years. This is the end of Part One. I hope you enjoyed it. We still have Treasuries and inflation to come. And based upon what I see with inflation, the old adage of history rhyming might be coming to an end. Until next time… Your Pulse on Crypto, Ben Lilly Also Published here P.S. - If you want to tune in to our weekly market update on crypto, head over to and subscribe. This way you can read the most timely takes as they go out. Also, be sure to check us out at to learn more about our services and team. https://jarvislabs.substack.com/ www.jarvis-labs.xyz

An Open Essay on the U.S. Federal Reserve, Treasuries, Cryptocurrencies and More: Part 1

It’s a captivating idea that can have you discussing unfathomable wealth creation in a short period of time. And the magnitude of how much change and growth is upon us is incomprehensible. To state it simply, it’s literally the biggest growth trend since the seventeenth century. A time when goldsmiths began practices that gave way to modern day banking practices such as the storage of wealth, providing loans, transferring money and providing bills of exchange that would lead to the development of bank checks. An idea of this size is called a grand supercycle. Now, you’re probably laughing a bit since the idea of bitcoin going through a supercycle is a relatively well known idea for those who follow the space. And many have called it “dead” in part from depressed prices. Then here I come in and try to “one up” the name by tossing the word “grand” in front of it. I assure you it’s not the case. The idea I’m presenting here is the idea that our entire society is about to enter the highest rate of growth ever. An enlightening period. And where it can all be found in a single chart. U.S. 30 year bond yields. It’s one I constantly stare at each day. Most recently I sat there trying to figure out an investment thesis over the next six to twelve months given the macro conditions taking place. Luckily, the answer was already on my screen and it evolved into something I’m calling Crypto’s Gospel. It’s the investment thesis everybody can get behind no matter where you stand. So join me as I take you through the thought process so if you ever find yourself speaking with somebody that has the means for change, this might give you that last tool in your bag to make a difference. The Correlation Now, when you take a look at the chart feel free to nit pick the similarities or lack there of in the comments section below. I agree there are some holes in this comparison. But one thing is for sure… Broadly speaking, you can’t disagree these two assets are moving similar to one another with a several day lag. So… what might be the story here? Well, bonds are a great barometer for the market and helpful gauge for future market conditions. And while it’s a tricky one to get right with so many variables at play, we need to have our vegetables from time to time. Which means occasionally digesting some rather unpleasant information is necessary for our health. Cue up the macro talk… Alright, in my daily newsletter we discussed some policy implications from the FED meeting that I’ll quickly cliff note here. Espresso The FED handed out a bit of yield to banks flush with cash. The way to think about this is the money printers in Washington created so much cash that banks ran out of places to stuff it… And they were so full they regurgitated the cash back to the FED. They did this by swapping their excess cash for collateral, aka treasuries, that the FED were now happy to dole out with some yield. This means banks are now getting yield on short-term treasuries and collateral unlike before. You can see the impact it had on treasury yields in the chart below with short dated treasuries getting a nice bump. Additionally, the FED hinted at a desire to raise rates sooner rather than later. While this is still a ways away, this lifted rates for other treasuries as well. So we sort of had two results from the FED meeting last week that played a role in yields. What’s interesting is not all durations or time to maturities followed suit. Here is what happened to the 30Yr yield. Dropped like a rock. When thinking about the amount of money which was printed in the economy, and how inflation is beginning to rear its head, how does this happen? After all, if inflation is expected to rise, the FED will need to raise rates to combat these pressures. And since higher rates mean stuffy language like discounting one’s future cash flow, this creates higher yields (lower bond prices). This all leads to the concept of higher rates slowing down economic growth. The Secret is Productivity But more importantly, and a less often talked about concern is higher rates slow down the “churn” of cash. Huh? I know there’s a lot coming at you at once here, so let’s break it up with an example… If a coin had a supply of 100 coins, and 30 of them are locked up… Meaning 70 are freely available. What happens when you add 50 coins to its supply? Besides there are now being 150 coins. The coins that are freely available go from 70 to 120. And as those coins move hands faster and faster in order to pay for goods and services, the coin loses value more and more. That’s because at first if nobody moved coins, the added supply would do nothing to prices. And it’s not until the coins begin to chase the same amount of goods and services that its price starts to drop. It’s price discovery in a way. More coins chasing after the same amount of goods means the price of goods go up relative to the coin. But consider this… if you slow down the movement of coins to where the locked amount or the amount that rarely transacts grows from 30 to 100… And now 50 coins are chasing the same goods and services, the price of the token is likely to rise again. This is the beauty of economic levers at work. It works because the good and services have less demand since the overall economy has a slowdown, which in turn means goods drop in price relative to coins to attract buyers… Giving the token holder more bank for their buck once again. It’s essentially the playbook for how the US fights inflation. The only thing here is we can’t expect this to happen again. Wait... what? The detour was relevant, I promise. The fact is, if you crank rates higher, the markets get whacked. We all seem to realize this lately. And it’s why we all grab popcorn before Powell talks now. But the thing here is markets more so than in prior times creates some feedback when it comes to spending more… And therefore, the growth of an economy (GDP). If your stock portfolio goes higher, you are more inclined to spend more. And also more easily access more cash. Which begs the question... How do you drop the market and raise rates at the same time? You can’t. It’ll crush economic growth in the long-term. The 30Yr yield reflects this current situation. That’s because 30Yr yields tend to rise when there is an that the economy will grow. expectation It’s how you get a normal looking yield curve at a time when everybody is anticipating inflation due to insane money supply growth. Now, bare with me here. I know this is getting a bit boring. But I promise it’ll help in getting my point across later. So before getting to it, let’s quickly reiterate… Too much cash in the system. Rates are rising in part because of the FED’s hand out of yield to banks and also because of inflationary forces beginning to percolate. And 30yr yields dropped below 2% for a hot minute since future growth looks muted. So why does this matter for bitcoin? And why might bitcoin be tracking the 30yr yield? The Supercycle Theory Well, I’m starting to form my own case for the supercycle theory. My version of it puts crypto into the limelight of economic growth discussions. And if crypto as a whole bands together and adopts it, I think the prospect of economic growth becomes not only real, but better than the highest economic forecasts out there. Consider this… How much of crypto is measured by GDP? I know stocks and equities are not measured in GDP… And therefore the crypto market cap should not be measured either. But there is an economy forming in crypto that feels like its off the record. Like a game of monopoly… Unmeasured GDP - h/t to Crypto Cobain in the UpOnly show with Raoul Pal, I got this non-existent GDP idea from him. What’s interesting is this untapped economy is more efficient than what is already measured in the normal economy. It’s in part why we are witnessing some of the smartest money in the world drop bucket loads of cash on young crypto projects. However, the truth is these stats are virtually unaccounted for. If crypto were able to grow and be accepted in the everyday commerce of life, the benefit for regulators, tax authorities, and lawmakers would be seen almost overnight. More growth, more taxes, happier citizens, and all those lovely feels. The way to see this is in viewing how many employees it takes to run Uniswap’s various smart contracts? None. Sure there were or still are about 16 or so employees… But now that those contracts exist in the wild, they can continue operating without any of those devs touching them. This makes crypto swaps via DEXs extremely efficient. The same operation in the legacy markets requires nearly 10,000 employees at traditional exchanges. Or the nearly 1,000 at Coinbase. Paving the way for these highly efficient businesses allows for the thousands or tens of thousands of employees in more traditional setups to build higher valued business, applications, and services by building on top of protocols similar to Uniswap. This is the catalyst for future growth. Any lawmaker that realizes this can instantly punch their ticket to a lifetime legislature seat if they lead the way towards clearer regulatory structures with low levels of red tape. If these structures are adequately framed, this freed up productivity will make its way into GDP growth. And for the lawmakers attached to these drafted proposals, this growth can be directly correlated to their name. Easily measured economic growth tied to their wisdom. The benefits for society are even greater. It can result in the highest levels of productivity in history. The third industrial revolution, exponential age, Web 3.0, The Fourth Turning, or whatever else you want to call it. These projects just starting to gain their footing in the cryptocurrency markets can be the foundation in which the rest of society’s technological advances will be built upon. And it’ll reduce costs as well as barriers to entry across the board. If you look throughout history, the growth in the number of businesses boosts GDP per cap. Crypto does this with thousands of businesses eagerly awaiting to set up shop in various jurisdictions around the world. It’s why Singapore has such a backlog of companies setting up shop right now regardless of necessary fees and hurdles. Opening up the economy to this type of growth is what the three piece suits walking the marble laid floors of the Federal Reserve call higher GDP per cap. It’s the pinnacle of goals for world leaders. Creme de la creme - Economic growth. And in order to get there the space needs clear regulatory structures. This allows cryptocurrencies to be brought into the conversation of economic growth along with other technological advances like AI and self-driving vehicles. And then act as a foundational layer - Web 3.0 - for a lot of technological developments taking place today. The main point here is crypto protocols enable services to happen with higher security and less cost. When there is nobody at the other end deciding the terms of your loan, and processing your payment… together with more security and less cost… this is measured efficiency. This enables business growth, more revenue, and similar prosperous outcomes for the economy. This might have seemed like a massive tangent to our conversation about 30Yr yields, but here's where I tie the two subjects together. When the broader population begins to see cryptocurrencies as part of the economic growth equation, the crypto space will ignite in a good way. What I expect to happen is a narrative change and almost enlightening moment as 30yr yields creep up again. This is in part the reflexivity argument. The positive feedback loop where higher prices means more belief. Only this time, as the belief in economic growth rises, 30Yr yields will rise along with it. And as these yields rise, the cryptocurrency market will follow along. This will then give way to crypto being a part of the conversation of creating more productivity in the economy. Reflexivity. To me, when that moment takes place the supercycle has arrived. It’s the moment cryptocurrencies go from a niche economy, to having a driver seat in the real economy. To date, bitcoin’s narrative is being driven by the inflation hedge rhetoric. This is accurate. But the missing piece here is cryptocurrencies as a whole enabling higher economic growth per person. Pairing these two together provides the investment thesis for bitcoin, layer one protocols, DeFi, NFTs, and so much more. Programmable money in a world of heightened inflation expectations that brings along with it higher economic growth per capita is where we are headed. This is not an argument just for bitcoin, ethereum, or DeFi. This is a crypto narrative. And I hope it helps band the crypto space together into a seamless front. Your Pulse on Crypto, Ben Lilly P.S. - To follow along join me on my blog: jarvislabs.substack.com

A Crypto Gospel: Supercycle Theory

The quietness of the market is either causing me to read the newsfeed more than normal, or my intuition is on to something. What I keep seeing are stories involving government and regulatory entities. The origin of these news items range from cities to nation-states to the world’s most powerful institutions. It’s like the entire globe is suddenly talking crypto. And it's filling up all my news feeds. The interesting part here is this is taking place NOT in the thick of a euphoric fervor. It’s during the quiet doldrums that often follow periods where the market gets chopped in half. Which causes me to ask why now? Is it all a coincidence? Is there something more afoot? Or am I really just killing time by digesting more news than normal and trying to come up with some sort of insight? To help me decide I went ahead and put together this issue. My hope is by the end of it I, and you, have a better understanding of the larger currents at play across the globe, and at the nexus of government and crypto. Now in order to keep this organized I’ll attempt to do it in a top-down manner. Meaning I’ll start with global entities, move to nation-4states, hit on a few non-categorical pieces like Facebook’s Diem that deserve their own space, and then tie it all together. Let’s see where it takes us. Global Entities First up is the top of the food chain. Global entities and organizations. These are the global institutions that can dictate the most widespread change. They can also end any progress with a flick of a pen. Let’s see what they have to say of late. First up… Bank of International Settlements (BIS) News out of BIS came out regarding how banks can handle cryptocurrencies. It came via their Basel Committee, a group that sets international banking standards. The committee proposed a 1:1 backing for cryptocurrencies (excluding stablecoins) held by the bank. Meaning the bank needed to have equal amount of capital reserves on hand. Thus, $100 in bitcoin required $100 in reserve. While the technicalities can fill the entire issue here, the key takeaway is the committee placed crypto in its own risk class outside of stocks and bonds. It’s essentially a green light for banks in holding crypto. And it might set the precedent to look at crypto as its own asset class. No more fitting a square peg into a round hole. Hopefully, others follow suit. World Economic Forum (WEF) We next saw some news from the WEF. This group helps world leaders understand trends and how to deal with them. They regularly publish toolkits in dealing with various issues such as cryptocurrencies. The latest . toolkit issued from WEF was centered on DeFi I found it fascinating to see an entire publication from the WEF centered on DeFi. After going through the document it appeared the toolkit recommended softer approaches to emerging protocols. It even recommended that leaders engage with developers and perhaps draft new rules and guidelines around these innovative creations. My impression is the WEF is telling world leaders DeFi is here to stay. When we combine the news from BIS and WEF, it appears the global elite don’t expect crypto to go away. And they are now beginning to work with it. To me, this means we can ignore a lot of the empty comments and media one liners from leaders who say crypto is nothing. Because if it were nothing, you wouldn’t see things like this created or discussed. Now it just comes down to individual countries and nations to develop their own rules and forms of adoption. So what are some of these countries saying? Nation-States Moving from global institutions down to nation-states let’s start with the groundbreaking news out of the smallest country in Central America. El Salvador Bitcoin is legal tender in El Salvador. (Mic drop.) You likely read this all over the crypto-sphere last week. It’s a monumental move for bitcoin as this makes the Latin American country the first in the world to recognize the king of crypto as money. It’s a new frontier for bitcoin and crypto as a whole. Very exciting. And when looking at what the leaders of the world are thinking in regards to this event, we can take a look at a note from JPMorgan. In a recent letter to their clients on the matter, they stated, [it might get] complicated if this is the beginning of a broader trend among similarly situated, smaller nations. Which is to say the bank believes this could escalate into a situation where smaller nations begin to adopt multi-currency systems using bitcoin, similar to what El Salvador is doing. Well, we likely won’t need to wait too long as those concerns are already starting to manifest. For example, Tanzania’s president called on its central bank to start preparing to facilitate for cryptocurrencies widespread use. Strong words from the leader. And while this was the strongest reaction so far... it wasn’t the only. We also witnessed some chatter from Nigeria, Brazil, Paraguay and Panama. Each showing varying levels of interest or praising El Salvador’s move. Now, the concern here in bitcoin being legal tender is that it might be a trend that starts via one country, but soon progresses into a trend with a dozen countries. Which sets up an interesting dynamic to watch unfold in the coming months. Particularly with the IMF. You see, El Salvador is in talks with the IMF for a $1 billion funding program to help make up its $3.2 billion budget deficit for 2021. With this recent move, El Salvador might soon be labeled as too risky or having loose monetary policy. And it could lead to some negative blowbacks by various global institutions that make up the established world order. To them, change is likely not a positive, particularly when it comes to the nation’s currency. But consider something else here. Consider what happens when you upset the little guy… If there is in fact blowback, how does El Salvador respond? Or better yet, how do other countries respond that appear to be backing their recent move? Do they band together? This is to say that maybe the IMF will try to avoid any potential outcome that creates a united enemy. We’ll see, but no matter what, this move by El Salvador has started a game of chess with the world order just as they appear to be warming up to it. India India decided recently to ditch the bitcoin ban agenda, and looks to create an asset classification. This is similar to what we saw from the global entities. For some background, a lot of what gets reported on India tends to be from unknown sources, which ends up causing a lot of FUD in the market. This was seen just a few months ago. with Ketan Surana, a member of the Internet and Mobile Association of India. The quote is, Well the latest news out of India looks to be quoted “We can definitely say that the new committee which is working on cryptocurrencies is very optimistic on cryptocurrency regulation and legislation.” This seems like a positive. And it comes after the government told banks to stop some of their negative actions against crypto. Note, anything coming out of India will take substantial time, likely on the magnitude of years. So change isn’t coming from the country this year. But if we ignore the time aspect for a moment, this might help clear up some of the FUD that occasionally makes its way into the media cycle. Thus, we have another type of news topic to ignore for the time being. And we can also see if other countries follow suit and begin to adopt similar narratives to India, which borrow from global entities. China This is one of the most unique situations in the world. China is pushing forward with the most ambitious roadmap of any country when it comes to its Central Bank Digital Currency (CBDC). Simultaneously, the government is cracking down on businesses taking part in the cryptocurrency space. This includes mining operations, custodians, exchanges, and any other business that touches the asset. The reason is likely connected to their agenda for their CBDC. By limiting the amount of capital moving into cryptocurrencies, they remove competition for their own currency - which on a side note was recently said to be able to transact on Ethereum in the future. This tells us the idea for China to have a global reserve currency does in include public blockchains in some capacity. To me, even though the currency won’t be issued on a public blockchain, the fact it might touch a public network is as big as adoption can get for the time being. That’s because if a global player like China looks at Ethereum as a way to move capital, then banning cryptocurrencies is not done to protect its citizens from speculative behaviors. It’s done for a first mover type of advantage. My belief is the government wants to onboard their citizens with their CBDC first, then control their movement into crypto. The idea here is to make sure the CBDC is the go to currency for its citizens, and not alternatives like BTC, USDt, or the Digital US Dollar. While the thinking here is more of an educated guess, it’s yet another story to watch unfold in the coming months. The news is not very positive coming out of China lately, but what we want to see evolve is how their stance towards crypto evolves once their CBDC gets released. United States The last nation to look at is the one that many in crypto have their eye on virtually 24/7. That’s because any decision out of the US tends to ripple through the markets, making it the one to watch when it comes to price action. Unfortunately, there is literally no further clarity whatsoever taking place here. If there is to be a bright side to this lack of clarity, the mere fact there are more regulators talking about crypto than ever before might be viewed as a positive. Over the last month we’ve seen comments from the SEC, IRS, Treasury, the FED, CFTC, OCC, Biden Administration, FDIC, Congress, and likely other three letter agencies I glossed over. And what the consensus seems to be among all these various agencies is simply if you play nice, don’t launder money, or do any shady shenanigans, then continue doing what you are doing. It’s frustrating for investors and project creators in the US as it’s pretty much a status quo of living in fear. And it’s what most have been doing since 2017. The issue is, how are we no further along in crypto than we were four years ago? What the heck is everybody waiting for? Why does the US continually kick the can down the road? Well, for brevity’s sake, let’s get on to what I believe every US regulator is waiting on… Digital US Dollar The Federal Reserve is set to talk this summer on what they have been researching when it comes to the Digital US dollar. My hunch is that whatever they reveal, it’ll be underwhelming. That’s because they are likely leaning on Facebook’s Diem to act as the test. Recall, Facebook’s Diem was originally Libra. It was a project-based out of Switzerland. The coin itself was a basket of currencies. Since then, the project has changed course. In fact, just a little over a week before Powell announced the upcoming discussion paper, Diem moved back to the United States. The same day it announced it was partnering with Silvergate Bank to issue the new US dollar-backed stablecoin, Diem USD. This is the same Facebook that is known to be lobbying the US government hard, and even poaching talent from their very own regulators. Which is to say, Facebook will not take no for an answer. And with such a flurry of events surrounding Diem, I had to take a deeper look as there is likely more to this sudden change in direction. After all, Silvergate is a Federal Reserve member bank. Which means checks and wire transfers pass through its payment system. So if Diem got issued as a pseudo Digital US dollar, the Federal Reserve will likely be able to monitor all its activity. Its a nice alternative for a bank, and using Diem will likely mean less pushback from US citizens. And we’re already seeing the government paving the way for Diem to take place via SWIFT. SWIFT is a financial messaging service and is better understood as a cross-border payment system. It is also known as a tool of the US government for getting entities outside its borders to comply with its wants by cutting off payments, issuing sanctions, or clogging up payment lines. Most recently SWIFT got swept up in crypto via Silvergate and the most popular crypto exchange. Silvergate customers until earlier this month were able to send payments to Binance using SWIFT. One of the reasons why a SWIFT / Silvergate would cut out Binance is due to the investigation involving the US Department of Justice (DoJ) and Internal Revenue Service. The two are probing the exchange under the pretense of a tax and money laundering investigation. If we are to be realistic about the investigation, the fact this is happening is understandable. Remember, all is good if you play by already established rules. What I find a bit odd here is the US DoJ. You might remember they and the CFTC (Commodity Futures Trading Commission) went after BitMEX for trading derivatives on bitcoin. After the event, liquidity left BitMEX in mass. Well, Binance trades derivatives as well. And if the DoJ and IRS find some of its citizens traded bitcoin futures on Binance, I wouldn’t be surprised to see these enforcement officials take the current probe to another level by taking some sort of measures against the largest derivatives exchange in the world. It’s likely the reason we saw SWIFT cutoff Binance. The currents at play are favoring entities that are working with Silvergate, the Federal Reserve payment network, and SWIFT. It is likely a cue of what might be coming to anybody not playing by its rules or agenda. Which leads us to the last piece of what regulators are waiting on. Tether You might remember our issue earlier this month called . In it, we talked about Tether, how USDT has its own economy in China, how KYC works with the stablecoin, and how it’s not back 1:1 with dollars. Fear Mongering Well, to hone in on that last aspect for a minute… Global entities are saying banks need to have 1:1 backing. And the BIS stated stablecoins need to be fully redeemable in addition to being regulated. We are also witnessing other countries possibly avoiding any competition to their own agendas in rolling out their own CBDCs, which in my opinion also includes the US. This makes Tether not the most attractive stablecoin in the eyes of regulators. Not to mention you already have Tether under the watchful eye of New York’s Attorney General. Tether is required to submit mandatory reporting periodically. If Tether has one hiccup, it won’t take much to cause the market to shift their stablecoin holdings from USDT to an asset that is favorable to US regulators like Diem or even USDC (has appropriate surveillance tools in place and already in the market). The key is timing. And when it comes to an issue like timing, I like to view it similar to herding cattle. You don’t want to spook the herd, but slowly guide it to where you ultimately want the cattle to end up. And do it without causing any harm to what is already built. Which leads me to my last few thoughts on the current affairs of crypto. Final Thoughts There are some pieces of crypto that are not looked upon kindly by regulators. The magic here is in how does one weed out the bad pieces or unwanted areas of crypto without causing harm to the rest of the space. To do so requires some subtle shifting and putting certain pieces in place. All the while not hindering the development taking place. Which means, once these pieces are finally in place (ie - Diem, USDC) the space will benefit from the clarity that follows. So what I am leaving you with today is a set of stories. And how these stories evolve over the coming year will dictate how crypto evolves. Do global regulators remain consistent in how they view crypto as an asset that will stay. If they change their tune, why? Does El Salvador cause the narrative to change? Do countries begin to soften up their view on crypto similar to the way India has done. And do these softer stances borrow from the narratives set by global entities? Will China embrace crypto once they release their CBDC? Is the US awaiting for Diem, and if so, what happens with Tether? These are stories that all stand on their own, but similarly influence one another. And I do not think it’s a coincidence we are seeing a flurry of news hitting our feeds. Things are coming to a head. The needed pieces are almost in place, it just takes a bit more time. This means regulators can begin to offer subtle guidance in the meantime in order to not disrupt innovation. So let’s continue to track these stories to ensure we don’t miss any opportunities or take unnecessary risks in the market. Your Pulse on Crypto, Ben Lilly P.S. - Did you enjoy this and want to follow along on insights delivered daily? Be sure to , the newsletter where this story was first published. subscribe to Espresso Also published on: https://jarvislabs.substack.com/p/state-of-affairs

The Crypto State of Affairs in 2021

(This post was originally posted to my personal blog ) Controlling modals of all kinds (informational modals, custom alerts, image lightboxes, …) across an application is a frustrating problem that we all face when developing apps. When the app is simple, it’s sufficient to place a single component and have a state to toggle it on/off. However, things get complicated when you have different places to present the same modal or different kinds of modal to display in your app. <Modal /> There have been a couple of design solutions for this such as using the React Context API ( is an example) or manipulating ’s (as described in ). In this post, I would like to share my own design for this problem using a singleton component. What we will achieve from this solution is an imperative API like: this blog react-navigation StackNavigator this blog post <Modal /> SomeModal.show(title, content, { ...options })SomeModal.hide() (Note that this is kinda similar to guide ) react-navigation Navigating without the navigation prop is a react component and it is rendered in the top-level like this: <SomeModal /> class App extends Component {  render() {    // Container is your own wrapper component    return (      <Container>        {/* Your real app */}        <SomeModal />      </Container />    );  }} So how does knows what to do when we call and methods? First, we will define component methods and , then in its constructor, we save its reference to a static property of by using (since in the constructor will be the reference to the rendered component). That reference will have access to all component methods of . We now can make two static methods and use it: SomeModal show() hide() __show() __hide() SomeModal this this SomeModal show(), hide() class SomeModal extends Component {  static show() {    SomeModal.__singletonRef.__show();  }  static hide() {    SomeModal.__singletonRef.__hide();  }  constructor(props) {    super(props);    SomeModal.__singletonRef = this;  }  __show() { ... }  __hide() { ... }} When doing this singleton component, we need to make sure we do not use this component twice in the app, since the reference will not be correct. This is the best fit, in my opinion, when we have a defined modal style across the app and we want to show and hide it everywhere we want to. This also allows showing and hiding modal outside of a React component. What if you have different kinds of modals? In my own project, for every kind of modal, I make a separate React component and render it to the top level like this: <Container>  {...}  <FeedbackModal />  <LinkPreviewModal />  <AlertModal /></Container /> Hope you found this way of working with modals useful. Please let me know your thoughts! Happy hacking!

Managing React modals with singleton component design

August 29th, 1997. That’s the day Skynet became self-aware, and the day it decided to attack humanity. Of course, that was in Terminator 2. A movie. Nothing close to real life. While it’s a classic, it’s guilty of emphasizing the negative consequences of technology that most humans couldn’t understand at the time. Things are different now. Artificial Intelligence (AI) has made unprecedented advancements in recent years: think self-driving cars, voice assistants, and smart home systems. And none of these things are “bad” or “evil.” Examples like from Marvel's series and Samantha from the film come to mind. Jarvis Iron Man Her While these AI systems have been glamorized and personified in Hollywood, it begs the question: are we on the cusp of seeing such advanced AI in our lifetime? Let’s get into the possibility and implications of such a reality. The Evolution of AI: A Historical Perspective AI has been touted as the next big thing for decades. The term 'Artificial Intelligence' was first used in the 50s by at a conference in Hanover, New Hampshire. computer scientist John McCarthy Since then, researchers have made impressive strides in developing AI systems capable of complex tasks and even learning from experience. From Turing's Test to Deep Learning In 1950, computer scientist Alan Turing proposed a test to determine if a machine could exhibit intelligent behavior equivalent to a human's. This became known as the “ ” and remains a benchmark for measuring AI. Turing Test Deep learning - a subset of - took center stage in 2012 with an AI system that could recognize cats in YouTube videos. machine learning The breakthrough sparked the current deep learning revolution, leading to natural language processing, image recognition, and autonomous decision-making advancements. Current State of AI: How Close Are We? Today's AI systems can perform a wide range of tasks. They can play complex games like chess and automate routine tasks in industries such as finance and healthcare. While these systems are impressive, they still fall short in terms of general intelligence - the ability to adapt and learn new things. Narrow vs. General AI The current AI landscape is dominated by narrow AI. These are systems efficiently. designed to perform specific tasks It can be used for professional and entertainment purposes. Many marketers use automation tools to help them do more in less time. General AI, on the other hand, can reason and think like a human. It can learn and adapt. That’s what would be required to come up with a Jarvis or a Samantha. But achieving general AI is no easy feat. We’ll need to push through significant scientific ceilings in areas such as natural language processing, emotional intelligence, and creativity, among others. The Path to Advanced AI: Challenges and Considerations While we may be closer than ever to witnessing advanced AI systems like Jarvis and Samantha, significant challenges still need to be addressed. These include: AI systems rely on vast amounts of data to learn and make decisions. Any bias in the data can lead to unideal outcomes. Data bias: Advanced AI systems with general intelligence raise important questions about safety and ethical considerations. Who is responsible if an AI system makes a harmful decision? How can these systems be used ethically? Safety and ethics: If advanced AI systems become a reality, how will they interact with humans? Will they be able to empathize and understand human emotions? Human interaction: Jarvis and Samantha: Breaking Down the Fictional Reality The idea of AI systems like Jarvis and Samantha is not entirely out of the realm of possibility. We’ve already seen significant advancements in natural language processing and emotional intelligence in AI systems. Companies like are laying the groundwork today for what AI will be tomorrow. It’s essentially AI as a Service, allowing users to chat with AI based on fictional characters. Users can also create their own chat buddy. Moemate But what can we see from brands like this five or 10 years from now? There’s no way to know, but the results so far are stellar. It’s also important to note that Jarvis and Samantha are highly romanticized versions of AI. They have personalities and emotions modeled after humans. While it is possible that we may see AI systems, such as what Moemate has developed, with similar capabilities in our lifetime, they won’t be exactly the same as the characters depicted in movies. The Future of AI: What Can We Expect? The journey towards advanced AI, like Jarvis and Samantha, is still evolving. While we have made significant strides in AI technology, we have yet to completely bridge the gap between today's AI and tomorrow's possibilities. The challenges, primarily concerning data bias, ethical considerations, and human interaction, need attention. However, with the rapid pace of technological advancements, we remain optimistic. The focus should lie on nurturing an AI future that is beneficial and safe for humanity. While the prospect of seeing AI like Jarvis and Samantha remains an enticing possibility, these AI systems will blend our technological accomplishments and our shared human values when they do come to life. These values may be much different from what we saw in movies over a decade ago.

Will We See AI Like Jarvis and Samantha in Our Lifetime?

A Brief History of NLP Applications in the 21st Century “ ” Jarvis, you up? . For you, sir, always Tony Stark then goes on to give a series of commands to check and access his uber-cool Ironman suit. Jarvis even warns Mr. Stark that terabytes of calculations need to be done before an actual flight, only to be met with Tony’s cocky reply - “ ” Jarvis! Sometimes you gotta run before you can walk. Looks like a normal conversation with a personal assistant, doesn’t it? Except that Jarvis isn’t a human being but a smart AI assistant capable of holding seamless conversations and available at your service, always. If you want to reconstruct Jarvis in reality, Natural Language Processing (NLP) would be the first technology you would consider. A subfield of AI, NLP is a machine’s ability to understand human language in speech or text thereby narrowing the communication gap between man and machine. With enough virtual assistants like Siri and Cortana in the market already, are we ready to pull off a Jarvis? Let’s see how NLP was used over the decades and where it’s future course lies. Where did it all start? In his paper “Computing Machinery and Intelligence” published in 1950, Alan Turing poses a revolutionary question - “Can machines think?”. Though the 1940s saw glimpses of NLP in machine translation, the now-so-popular “ ” (also known as the imitation game), opened the possibilities of machines trying to comprehend the complex field of linguistics. Turing test To fulfill Turing’s criteria, a machine has to understand a human’s natural language in order to frame an intelligent human-like response. Many attempts were then initiated to pass the Turing test. ELIZA, SHRDLU, and PARRY were all precursors of interest in NLP. After a brief lull in NLP research, due to a halt in funding for NLP research by the Automatic Language Processing Advisory Committee (ALPAC) formed by the US government, natural language processing acquired a new dimension with the “statistical revolution” of the 1980s. To know the detailed history of NLP up to the 1990s, I suggest you go through an article by Dr. Hancox and another on dataversity . What we are more interested in here is the bloom in NLP since the 1980s, when application of statistical methods grew significantly. It is important to note that NLP may not necessarily be the backbone of all the technologies described in this post, but they do involve NLP tasks to a certain extent. The Era of Machine Translation You could call machine translation as the genesis of teaching machines our natural language. The ease with which Google Translate works now has a long history dating back to the 1930s when began to be filed for translating machines. patents Translations saw the light of day in 1954 when the was conducted by scientists to automatically translate more than 60 Russian sentences into English. In the , Statistical Machine Translation (SMT) stole the limelight off rule-based translation when researchers at IBM (Candide Project) worked on a large corpus of French and English texts found in the Canadian parliamentary debates reports, letting go of linguistic rules and calculated probabilities instead. Georgetown-IBM experiment late ’80s Meanwhile, the corpus-based translation or Example-based machine translation (EBMT) was also scrutinized by groups of researchers in Japan who worked with previously translated examples or corpora to translate new words. By , Systran’s machine translation was available in a few CompuServe chat forums and in 1997 one of the oldest online translators - BabelFish was launched by AltaVista (later taken over by Yahoo) that could do 36 pairs of language translation on a web browser. 1994 In 2001 syntax-based machine translation (translating syntactic units instead of words) was , followed by phrase-based machine translation between 2003 and 2005 that was later used to propel Google translate in 2006. Bing and Yandex launched their translators soon after in 2009 and 2011 respectively. proposed established While silent was carried out on using Recurrent Neural networks for MT, a decade since its inception, Google Translate a radical leap in commercial-scale translation by employing neural networks that were trained on very large datasets. research ushered Termed as the Google Neural Machine Translation system(GNMT), the engine behind Google Translate could confidently handle close to 10,000 language pairs in 2016. Ever since NMT came in the picture, hybrid approaches to translation are being experimented with to close the gaps in NMT with the benefits of phrasal or rule-based machine translation. Does this mail scream “SPAM”? The widespread use of E-mail as a form of digital communication to date comes with the classic problem of ham and spam. While IP blacklisting and email header inspection were initially used to identify spam, these methods when individually deployed were easily bypassed, thanks to IP and email spoofing. Thus arose the need for email content to be analyzed along with non-textual elements, signaling the entry of NLP to spot spam. In 1998, the was proposed followed by a commercially viable work on the same lines by Paul Graham ( ) in 2002. Bayesian approach to filtering junk e-Mail A Plan for Spam The Bayesian method, one of the earliest methods of statistical NLP, differed from usual text-based filters by its ability to automatically learn new words seen in probable spams and hence classified mails better with its expanding vocabulary. A word cloud of common words in spam Bayesian filters got better by scrutinizing mail header content, pairs of words, and phrases resulting in lower false positives and higher accuracy of spam identification and were implemented in many modern email clients. To further decrease the number of spams, Google in 2015 its usage of neural networks in its spam filters bringing the spam rate down to 0.1%. announced “Hey, Siri! What’s the weather like today?” An important movement in Natural Language Processing is IBM’s (1962) and later Harpy in the 1970s that stand as precursors for digital speech recognition. By the mid-1980s, the Hidden Markov Method (HMM) was found to be highly useful in modeling speech. Shoebox Though considerable success was shown by IBM’s Watson winning Jeopardy! In 2011, intelligent virtual assistants as we know them today gathered momentum when Apple out SIRI on iPhone 4S in the same year. And yes, Raj finally found the woman he can talk to in the 2012 Big Bang Theory series :) rolled Google Assistant (initially called Google Now) and Microsoft’s Cortana followed suite with Amazon introducing Alexa in 2014 sitting inside a smart speaker - Echo. To view a complete timeline of voice assistants, check out this cool infographic by voicebot.ai. AI voice assistants quickly found their way into a host of apps and devices, the google home for example, and Amazon selling over Alexa-enabled devices by Jan 2019. 100 million Chatterboxes to the rescue A close relative of virtual assistants is the AI-powered chatbots that carry out human-like conversations on messaging apps and online chats. The idea of machines trying to converse with humans was set off by chatbots like ELIZA(1966), PARRY (1972) and later by (1988) that even tried to crack jokes. With advancements in ML and Natural Language Processing, AI chatbots can now hold seamless conversations with users by understanding user intent and providing relevant inputs. Jabberwacky China-based company WeChat built a smart chatbot in 2009 but AI chatbots gained prominence in 2016 when Facebook it’s messenger platform to build chatbots and within 2 years active chatbots were developed on the platform. opened 300,000 Customer service has been immensely boosted by AI chatbots due to their 24/7 availability, the capability to address customer complaints and resolve simple questions. AI chatbots powered by NLP, are ready to take on of all customer service interactions worldwide by 2021. 15% Sniffing sentiments behind statements Going beyond mere text analysis, efforts were taken to identify the feelings behind highly unstructured content and their contextual significance. Early on classifying movie reviews found on the web as positive or negative using ML techniques (Naive Bayes, SVM, and MaxEnt) was done by Bo Pang et al in 2002. research A subsequent in 2004 by Bo Pang and Lillian Lee, discussed the radical effects of bringing “subjectivity” (how relevant is a particular word in voicing the author’s opinion) into classifying sentiments. These publications, along with Turney’s , kick-started a revolution of sorts in Natural Language Processing since growing language exchange on the web (social media, blogs, newsfeeds) was mined to understand people’s opinions. paper contributions Also commonly referred to as opinion mining, sentiment analysis was adopted by information giants like Thomson-Reuters who their inhouse NewsScope sentiment engine to measure news sentiment of thousands of companies for improved trading decisions. developed By 2009, Google implemented changes to its search engine based on research and patents filed by them that explored the concept of aspect-based sentiment analysis to summarize reviews based on sentiment. papers Soon in 2010, Twitter feeds were to gauge public opinion during the British elections, an example being a company called Linguamtics that scrutinized over 130,000 Twitter accounts. analyzed 2010 British election tweets analysis Though people were highly skeptical of the analysis and its outcomes, it got the tech world excited enough to it out on the 2012 US Presidential elections as well. In 2015, it was that industry leaders like Intel and IBM used sentiment analysis to decipher their employees’ emotions. try reported Emoticons, sarcasm, and language complexities challenge the field of sentiment analysis, yet companies have started rapidly investing in understanding public sentiment to amplify customer experience and glean consumer insights. So where’s NLP heading to these days? Apart from the major applications of NLP mentioned in this article, topic modeling, document summarization, and character recognition are a few other problems solved by natural language processing. The current Natural Language Processing scenario is strongly dominated by . Once driven by recurrent neural networks (RNN), deep learning-based NLP is now experimenting with that have been useful with sentiment analysis, question-answering systems, and machine translation to a certain extent. Deep learning models are also used to speed up or consolidate standard NLP tasks of POS tagging, parsing, and named-entity recognition. Deep learning Convolutional Neural Networks(CNN) proved Until transfer learning proves its stance, deep learning models require large data sets which we aggregate for you at . Xtract.io As the excitement around NLP mounts, maybe a decade from now JARVIS could be a reality and you can go and save the world, . perhaps

How Far Are We From a Real World Jarvis? 

Beyond a discussion of whether or not is true AI, and guessing at whether Intel’s new neural chips would be better off with the more volumetric structure of an ‘ ’ ( ), what really sets Jarvis apart is persistence. Jarvis Infinity Stone my bet is they would be Not just conversational persistence, mind you, which is one of the biggest things lacking in today’s assistant, but also persistence in time and space. Consciousness has a funny way of being able to derive meaning, or at least usable data, as it navigates environments, and is then able to communicate that meaning in various ways and highlight various aspects of it, and even evolve its own understanding of it according to an ongoing conversational exchange. It also has a knack for developing contextual usefulness, which is akin to figuring out user stories according to environments. Sometimes it can do all that in real time: develop multidimensional meaning and applicable pathways, and even figure out ways to replicate and enhance their benefits according to context and environment. Jarvis does this. He can surf multiple virtual pathways according to domain of knowledge and related taxonomies, according to relatedness of time, space or narrative (a better, and far more encompassing word would be consciousness) clues, and according to conversational requirements, and often according to all three at once. This is what is actually happening when Iron Man requests some specific bit of knowledge and seconds later, out comes an applicable story. It may not always be perfectly useful, but it is an enrichment upon the currently available data, both in breadth and in vision (interpretation). Now, vision as such might require human level AI, but maybe it doesn’t. Maybe one way of pulling it off is through simple and truthful summarizing of interrelated knowledge according to weight of relatedness to clues within the initial conversation/demand, but especially weight of clues available in virtual knowledge. Interpreting human sentiment and being able to highlight the essential conclusions of human expression, be it article length or tweet length, would go a long way to adding depth to an assistant’s knowledge based answers. In short, I don’t expect AI assistants to be able to survey battlefield conditions in full spherical flow, pilot a suit, and at the same time live-process both a conversation and endless virtual information and (therein held) human consciousness landscapes, but I do want them to take advantage of their superior speed. I can google cooking recipes, and I can do so faster than any assistant. I can also input them into a knowledge management system far better than I would be able to do through voice. Spacially planning my to-do’s in Trello , or a simple text file — that is, adding and structuring them visually — is similarly a far quicker and more intelligent flow than doing it verbally. Perhaps such things would be improved by a persistent conversation flow, wherein an assistant would already knew where I want specific to-do’s placed, etcetera, but that is a matter of ease, not accomplishment. Where an AI’s speed would shine, even current AI, is in navigating vast amounts of data, nimbly maneuvering through entire realms of semantic interrelatedness, social graphs, and systems of meaning derived from human expression on a specific topic, and returning an applied summary of such. Even putting on a tourist’s hat, I want to be able to ask about the most scenic routes around the Tour Eiffel, and get not a link to an article, but a conversational reply giving me several alternatives, with details and highlighting benefits and problems, etcetera. Additionally, having conversational persistence, I would be able to request the assistant to display a written replica of a specific reply, or a list of sources for the information, or a list of the options already discussed, and then the display of a specific option (in this case a route) on a map. These are merely quick notes on the topic. I know AI designed for writing is already capable of accomplishing most of this, but the computing resources and time needed to pull off their neat tricks may not comfortably fit into the confines of a user device. Its probably more than my tablet can currently conjure. But then again, maybe it’s also a problem of design. More can currently be done to move AI assistants towards conversational persistence, systematized memory and more agile space-time intelligence.

What Does Iron Man’s Jarvis Have That Your your AI Assistant Doesn’t?

5 AI Tools for Effortless Content Creation: A 2024 Guide

After my latest post about how to build your own RAG and run it locally, today, we're taking it a step further by not only implementing the conversational abilities of large language models but also adding listening and speaking capabilities. The idea is straightforward: we are going to create a voice assistant reminiscent of Jarvis or Friday from the iconic Iron Man movies, which can operate offline on your computer. Since this is an introductory tutorial, I will implement it in Python and keep it simple enough for beginners. Lastly, I will provide some guidance on how to scale the application. Techstack First, you should set up a virtual Python environment. You have several options for this, including pyenv, virtualenv, poetry, and others that serve a similar purpose. Personally, I'll use Poetry for this tutorial due to my personal preferences. Here are several crucial libraries you'll need to install: : For a visually appealing console output. rich : A robust tool for speech-to-text conversion. openai-whisper : A cutting-edge library for text-to-speech synthesis, ensuring high-quality audio output. suno-bark : A straightforward library for interfacing with Large Language Models (LLMs). langchain , , and : Essential for audio recording and playback. sounddevice pyaudio speechrecognition For a detailed list of dependencies, refer to the link . here The most critical component here is the Large Language Model (LLM) backend, for which we will use Ollama. is widely recognized as a popular tool for running and serving LLMs offline. If Ollama is new to you, I recommend checking out my previous article on offline RAG: Basically, you just need to download the Ollama application, pull your preferred model, and run it. Ollama "Build Your Own RAG and Run It Locally: Langchain + Ollama + Streamlit." Architecture Okay, if everything has been set up, let's proceed to the next step. Below is the overall architecture of our application, which fundamentally comprises 3 main components: : Utilizing , we convert spoken language into text. Whisper's training on diverse datasets ensures its proficiency across various languages and dialects. Speech Recognition OpenAI's Whisper : For the conversational capabilities, we'll employ the Langchain interface for the model, which is served using Ollama. This setup promises a seamless and engaging conversational flow. Conversational Chain Llama-2 : The transformation of text to speech is achieved through , a state-of-the-art model from Suno AI, renowned for its lifelike speech production. Speech Synthesizer Bark The workflow is straightforward: record speech, transcribe to text, generate a response using an LLM, and vocalize the response using Bark. Sequence diagram for voice assistant with Whisper, Ollama, and Bark. Implementation The implementation begins with crafting a based on Bark, incorporating methods for synthesizing speech from text and handling longer text inputs seamlessly as follows: TextToSpeechService import nltk
import torch
import warnings
import numpy as np
from transformers import AutoProcessor, BarkModel

warnings.filterwarnings(
    "ignore",
    message="torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.",
)


class TextToSpeechService:
    def __init__(self, device: str = "cuda" if torch.cuda.is_available() else "cpu"):
        """
        Initializes the TextToSpeechService class.
        Args:
            device (str, optional): The device to be used for the model, either "cuda" if a GPU is available or "cpu".
            Defaults to "cuda" if available, otherwise "cpu".
        """
        self.device = device
        self.processor = AutoProcessor.from_pretrained("suno/bark-small")
        self.model = BarkModel.from_pretrained("suno/bark-small")
        self.model.to(self.device)

    def synthesize(self, text: str, voice_preset: str = "v2/en_speaker_1"):
        """
        Synthesizes audio from the given text using the specified voice preset.
        Args:
            text (str): The input text to be synthesized.
            voice_preset (str, optional): The voice preset to be used for the synthesis. Defaults to "v2/en_speaker_1".
        Returns:
            tuple: A tuple containing the sample rate and the generated audio array.
        """
        inputs = self.processor(text, voice_preset=voice_preset, return_tensors="pt")
        inputs = {k: v.to(self.device) for k, v in inputs.items()}

        with torch.no_grad():
            audio_array = self.model.generate(**inputs, pad_token_id=10000)

        audio_array = audio_array.cpu().numpy().squeeze()
        sample_rate = self.model.generation_config.sample_rate
        return sample_rate, audio_array

    def long_form_synthesize(self, text: str, voice_preset: str = "v2/en_speaker_1"):
        """
        Synthesizes audio from the given long-form text using the specified voice preset.
        Args:
            text (str): The input text to be synthesized.
            voice_preset (str, optional): The voice preset to be used for the synthesis. Defaults to "v2/en_speaker_1".
        Returns:
            tuple: A tuple containing the sample rate and the generated audio array.
        """
        pieces = []
        sentences = nltk.sent_tokenize(text)
        silence = np.zeros(int(0.25 * self.model.generation_config.sample_rate))

        for sent in sentences:
            sample_rate, audio_array = self.synthesize(sent, voice_preset)
            pieces += [audio_array, silence.copy()]

        return self.model.generation_config.sample_rate, np.concatenate(pieces) : The class takes an optional parameter, which specifies the device to be used for the model (either if a GPU is available, or ). It loads the Bark model and the corresponding processor from the pre-trained model. You can also use the large version by specifying for the model loader. Initialization ( __init__ ) device cuda cpu suno/bark-small suno/bark : This method takes a input and a parameter, which specifies the voice to be used for the synthesis. You can check out other value . It uses the to prepare the input text and the voice preset, and then generates the audio array using the method. The generated audio array is converted to a NumPy array and the sample rate is returned along with the audio array. Synthesize ( synthesize ) text voice_preset voice_preset here processor model.generate() : This method is used for synthesizing longer text inputs. It first tokenizes the input text into sentences using the function. For each sentence, it calls the method to generate the audio array. It then concatenates the generated audio arrays, with a short silence (0.25 seconds) added between each sentence. Long-form Synthesize ( long_form_synthesize ) nltk.sent_tokenize synthesize Now that we have the set up, we need to prepare the Ollama server for the large language model (LLM) serving. To do this, you'll need to follow these steps: TextToSpeechService : Run the following command to download the latest Llama-2 model from the Ollama repository: Pull the latest Llama-2 model ollama pull llama2 . : If the server is not yet started, execute the following command to start it: . Start the Ollama server ollama serve Once you've completed these steps, your application will be able to use the Ollama server and the Llama-2 model to generate responses to user input. Next, we'll move to the main application logic. First, we need to initialize the following components: : We'll use the Rich library to create a better interactive console for the user within the terminal. Rich Console : We'll initialize a Whisper speech recognition model which is a state-of-the-art open-source speech recognition system developed by OpenAI. We'll use the base English model ( ) for transcribing user input. Whisper Speech-to-Text base.en : We'll initialize a Bark text-to-speech synthesizer instance, which was implemented above. Bark Text-to-Speech : We'll use the built-in from the Langchain library which provides a template for managing the conversational flow. We'll configure it to use the Llama-2 language model with the Ollama backend. Conversational Chain ConversationalChain import time
import threading
import numpy as np
import whisper
import sounddevice as sd
from queue import Queue
from rich.console import Console
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationChain
from langchain.prompts import PromptTemplate
from langchain_community.llms import Ollama
from tts import TextToSpeechService

console = Console()
stt = whisper.load_model("base.en")
tts = TextToSpeechService()

template = """
You are a helpful and friendly AI assistant. You are polite, respectful, and aim to provide concise responses of less 
than 20 words.
The conversation transcript is as follows:
{history}
And here is the user's follow-up: {input}
Your response:
"""
PROMPT = PromptTemplate(input_variables=["history", "input"], template=template)
chain = ConversationChain(
    prompt=PROMPT,
    verbose=False,
    memory=ConversationBufferMemory(ai_prefix="Assistant:"),
    llm=Ollama(),
) Now, let's define the necessary functions: : This function runs in a separate thread to capture audio data from the user's microphone using the . The callback function is called whenever new audio data is available, and it puts the data into a for further processing. record_audio sounddevice.RawInputStream data_queue : This function utilizes the Whisper instance to transcribe the audio data from the into text. transcribe data_queue : This function feeds the current conversation context to the Llama-2 language model (via the Langchain ) and retrieves the generated text response. get_llm_response ConversationalChain : This function takes the audio waveform generated by the Bark text-to-speech engine and plays it back to the user using a sound playback library (e.g., ). play_audio sounddevice def record_audio(stop_event, data_queue):
    """
    Captures audio data from the user's microphone and adds it to a queue for further processing.
    Args:
        stop_event (threading.Event): An event that, when set, signals the function to stop recording.
        data_queue (queue.Queue): A queue to which the recorded audio data will be added.
    Returns:
        None
    """
    def callback(indata, frames, time, status):
        if status:
            console.print(status)
        data_queue.put(bytes(indata))

    with sd.RawInputStream(
        samplerate=16000, dtype="int16", channels=1, callback=callback
    ):
        while not stop_event.is_set():
            time.sleep(0.1)

def transcribe(audio_np: np.ndarray) -> str:
    """
    Transcribes the given audio data using the Whisper speech recognition model.
    Args:
        audio_np (numpy.ndarray): The audio data to be transcribed.
    Returns:
        str: The transcribed text.
    """
    result = stt.transcribe(audio_np, fp16=False)  # Set fp16=True if using a GPU
    text = result["text"].strip()
    return text

def get_llm_response(text: str) -> str:
    """
    Generates a response to the given text using the Llama-2 language model.
    Args:
        text (str): The input text to be processed.
    Returns:
        str: The generated response.
    """
    response = chain.predict(input=text)
    if response.startswith("Assistant:"):
        response = response[len("Assistant:") :].strip()
    return response

def play_audio(sample_rate, audio_array):
    """
    Plays the given audio data using the sounddevice library.
    Args:
        sample_rate (int): The sample rate of the audio data.
        audio_array (numpy.ndarray): The audio data to be played.
    Returns:
        None
    """
    sd.play(audio_array, sample_rate)
    sd.wait() Then, we define the main application loop. The main application loop guides the user through the conversational interaction as follows: The user is prompted to press Enter to start recording their input. Once the user presses Enter, the function is called in a separate thread to capture the user's audio input. record_audio When the user presses Enter again to stop the recording, the audio data is transcribed using the function. transcribe The transcribed text is then passed to the function, which generates a response using the Llama-2 language model. get_llm_response The generated response is printed to the console and played back to the user using the function. play_audio if __name__ == "__main__":
    console.print("[cyan]Assistant started! Press Ctrl+C to exit.")

    try:
        while True:
            console.input(
                "Press Enter to start recording, then press Enter again to stop."
            )

            data_queue = Queue()  # type: ignore[var-annotated]
            stop_event = threading.Event()
            recording_thread = threading.Thread(
                target=record_audio,
                args=(stop_event, data_queue),
            )
            recording_thread.start()

            input()
            stop_event.set()
            recording_thread.join()

            audio_data = b"".join(list(data_queue.queue))
            audio_np = (
                np.frombuffer(audio_data, dtype=np.int16).astype(np.float32) / 32768.0
            )

            if audio_np.size > 0:
                with console.status("Transcribing...", spinner="earth"):
                    text = transcribe(audio_np)
                console.print(f"[yellow]You: {text}")

                with console.status("Generating response...", spinner="earth"):
                    response = get_llm_response(text)
                    sample_rate, audio_array = tts.long_form_synthesize(response)

                console.print(f"[cyan]Assistant: {response}")
                play_audio(sample_rate, audio_array)
            else:
                console.print(
                    "[red]No audio recorded. Please ensure your microphone is working."
                )

    except KeyboardInterrupt:
        console.print("\n[red]Exiting...")

    console.print("[blue]Session ended.") Result Once everything is put down together, we can run the application as shown in the video above. The application runs quite slowly on my MacBook because the Bark model is large, even in its smaller version. Therefore, I have slightly sped up the video. For those with a CUDA-enabled computer, it might run faster. Here are the key features of our application: : Users can start and stop recording their voice input, and the assistant responds by playing back the generated audio. Voice-based interaction The assistant maintains the context of the conversation, enabling more coherent and relevant responses. The use of the Llama-2 language model allows the assistant to provide concise and focused responses. Conversational context: For those aiming to elevate this application to a production-ready status, the following enhancements are recommended: : Incorporate optimized versions of the models, such as whisper.cpp, llama.cpp, and bark.cpp, which are designed to boost performance, especially on lower-end computers. Performance Optimization : Implement a system that allows users to customize the bot's persona and prompt, enabling the creation of different types of assistants (e.g., personal, professional, or domain-specific). Customizable Bot Prompts : Develop a user-friendly GUI to enhance the overall user experience, making the application more accessible and visually appealing. Graphical User Interface (GUI) : Expand the application to support multimodal interactions, such as the ability to generate and display images, diagrams, or other visual content in addition to the voice-based responses. Multimodal Capabilities Finally, we have completed our simple voice assistant application, full code can be found at: . This combination of speech recognition, language modeling, and text-to-speech technologies demonstrates how we can build something that sounds difficult but can actually run on your computer. Let's enjoy coding, and don't forget to subscribe to so you don't miss the latest in AI and programming articles. https://github.com/vndee/local-talking-llm my blog Also published here

How to Build Your Own Voice Assistant and Run it Locally Using Whisper + Ollama + Bark

কীভাবে আপনার নিজের ভয়েস সহকারী তৈরি করবেন এবং হুইস্পার + ওল্লামা + বার্ক ব্যবহার করে স্থানীয়ভাবে এটি চালাবেন

So erstellen Sie Ihren eigenen Sprachassistenten und führen ihn lokal mit Whisper + Ollama + Bark aus

Cómo crear su propio asistente de voz y ejecutarlo localmente usando Whisper + Ollama + Bark

Comment créer votre propre assistant vocal et l'exécuter localement à l'aide de Whisper + Ollama + Bark

अपना खुद का वॉयस असिस्टेंट कैसे बनाएं और इसे व्हिस्पर + ओलामा + बार्क का उपयोग करके स्थानीय रूप से कैसे चलाएं

独自の音声アシスタントを構築し、Whisper + Ollama + Bark を使用してローカルで実行する方法

나만의 음성 어시스턴트를 구축하고 Whisper + Ollama + Bark를 사용하여 로컬에서 실행하는 방법

Como construir seu próprio assistente de voz e executá-lo localmente usando Whisper + Ollama + Bark

Как создать собственный голосовой помощник и запустить его локально с помощью Whisper + Ollama + Bark

Kendi Sesli Asistanınızı Oluşturma ve Whisper + Ollama + Bark Kullanarak Yerel Olarak Çalıştırma

Cách xây dựng trợ lý giọng nói của riêng bạn và chạy nó cục bộ bằng Whisper + Ollama + Bark

如何构建您自己的语音助手并使用 Whisper + Ollama + Bark 在本地运行

Artificial intelligence gives humanity tremendous opportunities. With that, great risks. We celebrate online translators, amicable chatbots, well-drawing neural networks, and robot vacuum cleaners. We jubilate everything that makes life easier — not to mention scientific achievements. That said, we seem to be afraid of AI. But why? Is it even justified? Let's find out. First Steps in AI The first research on AI began in the 1930s. At that time, AI was already deemed a computer program able to perform tasks and learn. In 1927, an engineer of the Westinghouse Electric and Manufacturing Company, engineered the first humanoid robot, Herbert Televox. Roy Wensley It was programmed to answer calls and execute simple commands. Imagine you call home and ask Televox whether the lights are turned off! Or you can request to preheat dinner in the oven, and the robot will turn it on remotely. All that in 1927! Rise of AI: Dartmouth Conference The term artificial intelligence was first used in 1956, at the , New Hampshire. At the conference, scientists discussed many areas of AI application, from vision, learning, and research to language, gaming, and human interactions with robots. Dartmouth Conference in Hanover Since then, the golden age of AI has started. Artificial intelligence has become the official field of science. AI robots could solve specific tasks. For more accomplished versions, a bigger computing capacity was necessary. Not So Positive AI: Intelligence Explosion The number of robots — from a vacuum cleaner to a smart kettle — used in daily life was growing. Researchers tried to predict the further flow of things. In 1966, British mathematician Good in his work ‘Speculations Concerning the First Ultraintelligent Machine’ formulated the principle of technological singularity, or intelligence explosion. Irving John According to this phenomenon, “ultraintelligent machine could design even better machines; there would then unquestionably be the ‘intelligence explosion,’ and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make.” This work caused a great discussion in society and led to ethical concerns: Is it safe to interact with machines? Where are the boundaries of interactions? Who controls robots? Trade-Offs: Asimov’s Laws of Robotics One of the most famous solutions to the problem was proposed by the American writer and professor of biochemistry in 1942. In Asimov’s story ‘Runaround,’ he outlined the three Laws of Robotics: Isaac Asimov A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. But, many years later, these postulates remained fiction and became the ground for future doubts, fears, and frightening scenarios. Where Does the Fear Come From? In the 1990s, neural networks became a real breakthrough. They could collect data, form experiences, analyze the received information, and draw conclusions at much higher speeds. Since there were more precise and less provocative definitions for modern technologies, the phrase artificial intelligence lost its relevance. John McCarthy, an American computer and cognitive scientist, said, ‘As soon as it works, no one calls it AI anymore." It is human nature to be scared of the new — the law of survival in action. Artificial intelligence provokes a defensive reaction. But it’s not the AI itself that frightens us. It’s the fear of the unknown. Many do not understand artificial intelligence, its processes and its consequences. People are also scared lest artificial intelligence replace them at work. Here also comes media. In movies, books, and media, AI is often an evil genius who wants to destroy humanity. At the same time, there are many stories of human interaction with technology: Self-realization through a robot (the movie ‘Real Steel’); Complete suppression of human will by a machine (the picture ‘Upgrade’); Use of technology for good (Iron Man's virtual assistant Jarvis); Capture of the world by intelligent machines (the movie ‘Terminator’ or the ‘Matrix’ trilogy). Smart machines appeared in fantasy long before they did in real. The word ‘robot’ was first used in the science-fiction play R.U.R. by in 1920. Čapek’s robots were very close to the modern perception of androids. Karel Čapek People view AI as a living organism capable of evoking feelings, if not to replace a person. We can see that in the picture ‘Her.’ Moreover, we no longer perceive technologies like Siri, Mercedes, or chatbots as AI. However advanced, machines do not have a nervous system. Even a perfectly accurate emulation of the human brain in an electronic version cannot make a robot feel and empathize. Mikalai Dzisko*, Senior Editor at , says:* Ocean Power ‘Journalists also distort the image of neural networks and artificial intelligence. Instead of educating their audience and showing them how to use the neural network, they mouse around for intolerance and unethical behavior in ChatGPT. Will it write the f-word? When will it offend a user? Is it possible to bring it into conflict? How many genders does the neural network admit? These questions blur the reputation and essence of the technology. Will the Fear Of AI Disappear? Despite widespread fear, some movies, books, and people support a positive reputation of neural networks: In the 1990’s sci-fi picture ‘Total Recall,’ AI helps the main character to investigate crimes on Mars; In the Mass Effect computer games, you can’t win without AI; In the science fiction series ‘Black Mirror,’ artificial intelligence is sometimes represented as a positive character: in one of the episodes, it helps people find the perfect match — the utopian Tinder of the future. Artificial intelligence is often mentioned in gender studies. One of the most famous feminists of the last century, Donna Haraway, believed that cyborgization would help overcome binary, with all the ensuing problems and stereotypes. Today, human rights defenders are moving on. The non-profit organization raises the questions of people who can already be called cyborgs. Cyborg Foundation Its co-founder, artist Neil Harbisson, calls himself the first cyborg in the world because he implanted a special antenna in his skull that converts light waves into sound. ‘Should they have a license to use the technology? Their international law? What should cyborg insurance look like? Who will protect the civil liberties and inviolability of cyborg bodies?’ asks Eliezer Yudkowsky, co-founder and research fellow at the Machine Intelligence Research  Institute Stephen Hawking made a huge contribution to the popularization of the thesis ‘Technology is man's best friend.’ The famous physicist and cosmologist publicly spoke in favor of research and the development of artificial intelligence. Hawking believed that, with proper use, AI can help people solve many global problems. By personal example, he showed that a serious illness is no longer an obstacle to active scientific and public activity the moment you let technology into your life and body. Transhumanism VS. Posthumanism Transhumanists preach that the main task of a person is to improve the quality of their own life and overcome illness and death. Technology should be the best helper in this process. Cyborgization, regenerative medicine, various digital practices — virtual reality, metaverses — all this should make a person happy. Sounds defiant, doesn’t it? On the other hand, technology has been simplifying our lives for a long time. We just do not notice it: cars, microwaves, phones, the Internet, search engines, and so on. There is also a more radical approach, posthumanism. This philosophy is based on the consideration that man has reached the peak of their evolution in the Darwinian understanding and, to reach the next level; it is necessary not just to apply the product of his intellectual activity, but to directly implement it into life: to implant a chip in the brain to improve vision or hearing. This can also include biohacking, improving health indicators with the help of medicines, a proper lifestyle, and, of course, technologies, such as chips that read blood pressure, temperature, and sugar levels. Not the treatment of diseases, but the upgrade of the body and its own capabilities. Three Whales of Transhumanism Both trends are mostly theoretical, but the three main ideas of transhumanism seem intriguing. The first is super longevity. The belief that life is finite, and even more so limited by a conditional century, is a cultural stereotype that forces a person to distribute events within the time allotted to a human: to get married before the age of 25, to give birth before 30, to reach success before 45. What if you imagine that life is 200, or even 250 years? How will our attitude towards the time change? To your own health and biological age? What if we don't rush, try to do more, and adjust our lifestyle to this idea? A second duty of a transhuman is to raise the standard of living by reducing stress. Stress has become the main disease of the 21st century, the primary source of other problems and illnesses. A high standard of living is not only material goods, it is primarily harmony with oneself. This is what transhumanism adherers suggest to focus on. The third thesis of transhumanism is as follows: a person must combine the abilities of the brain and technology. Disturbing, right? In fact, we have been delegating some brain functions to neural networks for a long time: memory, sensory organs, and biological clocks, to name a few. We get up on the alarm clock, we do not waste time trying to remember the actor's name, we trust the route suggested by the navigator — and we are unlikely to follow it on our own. The neural network helps eliminate some processes, change our needs, reduce the time for completing tasks, and manage to do the important ones. Is It Even Possible? In theory, it is possible to achieve superintelligence by combining the capabilities of the human brain and a computer. Special implants could stimulate brain function, allow you to access the Internet, and interact with computers or other implant carriers. Today, such technologies are used only for medical purposes. For example, scientists at have developed implants that can stimulate individual parts of the brain with the help of electrical impulses and thus help address depression or Parkinson's disease. Stanford University All these utopian developments can inspire or make you think about improving your life. In the fight against AI anxiety, you just need to start. The first piece of advice unanimously given by those who work with neural networks is to stop calling this technology artificial intelligence. Intelligence — the defining characteristic between a human and an animal — remains man's main strength. There are no competitors on the horizon so far.

AI: Friend or Foe? What's Behind Our Fear of Artificial Intelligence?

Large language models (LLMs), representing the new wave of cutting-edge generative AI, exhibit remarkable capabilities across a wide range of domains, from natural language processing to content generation. However, discussion around a persistent flaw has surfaced recently, plaguing developers and businesses alike—unsupported or illogical information, commonly called "hallucinations." To get around this, some AI application development teams are increasingly adopting Software as a Service (SaaS) tools that leverage the power of generative AI and adopting a human-in-the-loop workflow to achieve an increase in model accuracy. A recent study on the “ ” shed light on a straightforward trajectory for the incident response lifecycle. It underscores the importance of using a unified SaaS tool or platform that seamlessly integrates all incident management tools utilized by organizations. This tool should not only harness insights from human data but also leverage generative AI, enhancing operational efficiency and facilitating informed decision-making. State of DevOps Automation and AI In this piece, we will delve into the integration of human-in-the-loop (HITL) approaches within multi-agent AI systems to unlock the full potential of LLMs while mitigating their inherent limitations. The Vision of Multi-Agent LLMs We would love to get to a point where there is an AI ecosystem with various autonomous agents that complement each other's strengths, working in harmony with ongoing human involvement. The market suggests that hopefully we will reach this point soon. In 2023, the revenue of the worldwide ‘ ’ surpassed $4.8 billion, and it is anticipated to reach approximately $28.5 billion by 2028. Autonomous AI and Autonomous Agents Market The "human expert in the loop" model involves combining generative language models, logic-driven quality assurance systems, and specialist content models under human supervision. This selective combination ensures tailored solutions that contribute unique strengths toward a reliable and trustworthy output. The task characteristics of LLM-based AI automation agents introduce various nuances. Reading data is generally considered safer than writing data, which requires human-in-the-loop validation to avoid unintended consequences. As tasks become more complex, involving multiple agents or skills, aggregating their outputs becomes crucial for overall reliability and performance. However, the narrative expands beyond the individual capabilities of LLMs. The emergence of multi-agent LLMs signifies a leap forward in AI architecture. Structured as AI systems facilitating dynamic conversations between a collection of language models serving as the reasoning engine for AI agents, multi-agent LLMs offer a collaborative approach to problem-solving. Each agent possesses distinct strengths and weaknesses, akin to assembling a team with diverse expertise. Unlike traditional LLMs operating in isolation, multi-agent LLMs engage in natural language interactions with each other, autonomously working on complex tasks. The modular nature of this approach increases versatility, enabling developers to create reusable LLM components for custom applications. This adaptability makes multi-agent LLMs well-suited for intricate tasks like those in autonomous vehicles, supply chain management, or cyber risk detection and mitigation. Powerful Use Cases Unleashed by Multi-Agent LLMs The applications of multi-agent LLMs are far-reaching: Healthcare, Finance, and Education In the medical field, where accurate and timely information is critical, multi-agent LLMs provide on-demand expertise for diagnostics and treatment options, enhancing patient care. Financial institutions can leverage multi-agent LLMs to analyze market trends, assess investment strategies, and offer personalized financial advice, elevating services and customer satisfaction. Finally, multi-agent LLMs could transform education by providing students access to diverse subject matter experts, fostering personalized learning experiences, and driving academic growth. Personal Agents One of the most crucial uses of an AI agent would be a Jarvis-like personal assistant on electronic devices accomplishing tasks based on text, voice, or gesture commands. Multi-agent LLMs can revolutionize the concept of personal assistants by enabling seamless collaboration between different agents, enhancing the user experience, and expanding the range of tasks that can be efficiently handled. Software UX Software user experience can be enhanced manifold by implementing efficient agents. Instead of manually browsing and clicking buttons to get things done, such as ordering food, cab, shopping, etc. AI agents will automatically accomplish them based on voice commands. Multi-agent LLMs can be key in transforming software interactions, making them more intuitive and user-friendly by understanding natural language instructions and executing tasks seamlessly. However, these powerful applications come with challenges. Integration complexities, data sharing, and ensuring seamless coordination among agents are hurdles that must be overcome to unlock the full potential of multi-agent LLMs. Implementing Multi-Agent LLMs Effectively To harness the potential of multi-agent LLMs, organizations must address key implementation considerations. As outlined in , there are a number of challenges to implementation, mainly from difficulties in sharing data and the need to ensure smooth coordination among agents. a recent Gartner study Firstly, robust data pipelines are essential for optimal multi-agent LLM functioning. Seamless information sharing among agents ensures access to the most up-to-date and relevant data. Secondly, establishing efficient coordination mechanisms involves defining rules, protocols, and communication channels to ensure cohesive work towards common objectives, enhancing overall performance. And lastly, continuous monitoring and maintenance are vital for long-term success. Implementing monitoring tools, regular updates, and maintenance address issues, improve efficiency, and adapt to evolving requirements. Ethical Considerations in a Global Context The global impact of multi-agent LLMs extends beyond individual organizations and industries — the public uses them, too. Detecting and mitigating bias in data and algorithms is paramount to guarantee fair and unbiased responses. Transparency in decision-making processes is equally essential, requiring organizations to provide explanations for generating specific responses or recommendations to build trust. Additionally, establishing clear lines of accountability within multi-agent LLM systems is crucial, defining responsibility for the actions and decisions made by the agents. Collectively, these principles form a foundation for the responsible deployment of multi-agent LLMs in various contexts. Multi-agent LLMs require international collaboration to reap global benefits. Governments, organizations, and researchers must work together to establish ethical guidelines, invest in cybersecurity measures, foster research and innovation, and ensure that the benefits reach underserved communities worldwide. Luckily, as of 2020, those from both private and public sectors have . published well over 100 ethical guidelines Multi-agent LLMs represent a transformative step in AI evolution, unlocking powerful use cases across various industries. As we navigate the complexities and challenges, international collaboration and ethical considerations will shape the trajectory of these advanced AI systems, steering toward a future where human and machine intelligence collaboratively drive innovation and progress.

Unlocking Powerful Use Cases: How Multi-Agent LLMs Revolutionize AI Systems

Every year, in early January, tens of thousands of tech aficionados descend on Las Vegas for the . It’s an event best known for capturing the tech world’s current zeitgeist; 2023 was heavy on , while 2022 revolved around . Unsurprisingly, CES 2024 was all about AI. , , … this year had it all. Customer Electronic Show (CES) EVs and sustainability improved hardware and gaming setups AI for headsets AI for babies AI for pets Obviously, the press unofficially crowns winners and losers at the end of the event. There is a variety of opinions on the losers ( ), but indisputable; . Google seems to be getting a lot of votes the winner is it’s the Rabbit R1 In short, R1 is a compact, AI-powered handheld device designed to streamline daily tasks. You talk into it like you would a walkie-talkie (by pressing a side button), and get the Artificial Intelligence to tasks historically done through apps. This includes, as per the demo, planning itineraries, ordering food, or booking a taxi. Speak, then validate with a click, without browsing / tinkering. The device is equipped with a 360° camera that allows it to “see” and better answer questions regarding the physical world. Finally a piece of software accessible via a PC allows its users to “teach” it repeatable tricks to save even more time. autonomously execute https://www.youtube.com/watch?v=22wlLy7hKP4&embedable=true Lauded for its efficiency and simplicity, the R1 is aiming to be a to our Tony Stark. A perfect assistant. , the company making it, has already pre-sold ! J.A.R.V.I.S. Rabbit at least 30,000 units Sadly, this ( ) cute piece of tech is severely over-hyped and doomed to fail. very The Rabbit R1: the Most Over-Hyped Project in the Tech World Right Now In , Rabbit’s founder contends that Amazon’s Alexa, Apple’s Siri and Microsoft’s Cortana may understand us, but are unable to complete actions. They’re a way to use Google verbally. Hardly . He portrays the Rabbit R1 as an AI companion able to do much more, from planning a complex holiday in a few sentence to accessing your favorite artists’ concert dates. And that’s a great idea! his keynote just assistants But it’s just idea. One idea . one that could easily and entirely fit inside a single app created by a larger player Rabbit may be the first out the gate (and props to them for that), but every single device maker has had the same plan for the past 18 months. . Microsoft recently announced , specifically for its Copilot AI assistant. , meanwhile, has been a Mac staple for a while now, and is also accessible at a click of a button. Even I’ve had this idea a year ago it is adding a new key to PC keyboards for the first time since 1994 Raycast The only reason we don’t have more examples is because natively integrating app in a vast universe of products and services takes time. This is however where the value will come from: a flywheel effect wherein an actor like Apple, being already connected to all your apps (and location) via iPhone iCloud data is able to better assist you by better knowing you. yet one and Having built a whole device for a functionality literally every smartphone is going to introduce and improve on in a few months, Rabbit is essentially just another . The only reason this is a separate device instead of an app is because the company needs to have unfettered access to a camera and microphone without Apple or Google stopping them. AI wrapper Which brings us to another obvious challenge: most individuals prefer carrying or using a single device, with their phones serving as the ideal choice ( ). My iPhone will soon do what the Rabbit does… and already does so much more, today. Should you purchase a Rabbit, it will absolutely stay home when you go out. when was the last time you bought a flashlight Then we get to the R1’s party trick. The device can connect to one’s apps (Spotify, Uber, Doordash…) to take actions on one’s behalf without having to visit said apps (with the user’s permission / login via a PC). This is neat… but what if tech giants like Meta or Google or Apple refuse to play ball and make integrations with the R1’s software more complex? And what if the apps no longer want to be accessed by an AI? This type of device could seriously hurt their ad or pull-through revenues. ; how will this be any different? We’ve already seen lawsuits aimed at OpenAI / ChatGPT Strategically, this is a massive vulnerability. On the other side of the spectrum, not only do large actors like Apple and Microsoft have a breadth of offers to concatenate, they also have the power to negotiate with apps makers to provide their services via AI… and pay them for it. Which leads us to the device’s economics. , an unusually low one-time (!) price for such a product. Again, good for them, … the R1 does not use fully local AI, so you’re using their cloud. And running a cloud is expensive ( ). It’s so expensive, in fact, that a one-time purchase rather than a subscription model makes little sense, and is unlikely to last. This means the company is likely to look for new growth avenues. If I were a betting man, I’d put my money on all the “privacy-first” promises disappearing when they realise data can be sold at a good price. The Rabbit R1 is priced at $199 but 110k$ a day for ChatGPT A local AI would have been a game changer. In the future, every device, such as TVs, cars, phones, and individual applications will come equipped with built-in natural language models. These models will be small, local and limited to their purpose. What we got instead is a Pokedex connected to the Cloud. Other issues, in bullet-points, because you only have so much time to study bad tech in details During CES, Google announced it was ’s “brain”. That obsolete functions include “Checking personal travel itineraries by voice” and “making a reservation”… which were both in the R1 demo! Google has enough data to know what it’s doing: no one is using an AI assistant to do these things. removing “under-utilized features” from its AI assistant The hardware looks cool, but is pretty dreadful in terms of wasted space; around 20% of the device is empty / unused. The keyboard works poorly (allegedly), which is unfortunate : people text input because they have needs that are detailed and complex, which are not easily expressed or recorded through voice. need AI hallucinations are not addressed. Since this is an AI and nothing else, how does one check the veracity of its statements? It’s supposedly very fast (though the problem with ChatGPT was never its speed). But in the demo, I wasn’t convinced it was faster than ChatGPT4. What if you're left-handed? I could be wrong about the R1. Its first batch sold out on day One and pre-orders are through the roof. It has generated considerable consumer interest. They are the first(ish) on the market with such a gadget, and though they will be trounced by Siri 2.0, the team should be proud. It definitely seems like the early days of a tech wave. There were all kinds of weird off-the-wall designs and ideas for phones at the beginning of the smartphone wave as companies were exploring to try and find niches and see what was possible. It’s kinda fun and I’m excited to see what other weird things people come up with. Good luck out there. Also appears . here

The Rabbit R1 May Be the Most Over-Hyped Tech Gadget at CES 2024

CES 2024-এ খরগোশ R1 সবচেয়ে বেশি হাইপড টেক গ্যাজেট হতে পারে

Der Rabbit R1 ist möglicherweise das am meisten gehypte Tech-Gadget auf der CES 2024

El Rabbit R1 puede ser el dispositivo tecnológico más publicitado en CES 2024

Le Rabbit R1 pourrait être le gadget technologique le plus surfait du CES 2024

रैबिट R1 CES 2024 में सबसे अधिक चर्चित टेक गैजेट हो सकता है

Rabbit R1 は CES 2024 で最も誇張されすぎたテクノロジー ガジェットになる可能性があります

Rabbit R1은 CES 2024에서 가장 과장된 기술 장치일 수 있습니다.

O Rabbit R1 pode ser o gadget tecnológico mais badalado na CES 2024

Rabbit R1 может стать самым разрекламированным техническим гаджетом на CES 2024

Tavşan R1, CES 2024'te En Çok Abartılan Teknoloji Gadget'ı Olabilir

Rabbit R1 có thể là tiện ích công nghệ được thổi phồng quá mức tại CES 2024

Rabbit R1 可能是 CES 2024 上最被炒作的科技产品

This paper is available on arxiv under CC 4.0 license. Authors: (1) Berrenur Saylam; (2) Ozlem Durmaz ¨ ˙Incel. Table of Links Highlights Abstract & Introduction Federated Learning: Collaborative and Privacy-Preserving Model Training State of the Art Federated Learning Software Platforms and Testbeds Sensors and Edge-Sensing Devices Federated Learning Applications on Sensing Devices Discussion Conclusions, Acknowledgment, and References 7. Conclusion This paper has provided an overview of the current state-of-the-art FL methodologies applied to sensing devices, focusing on IoT sensors, mobile devices, and wearables. By examining the challenges and opportunities in applying FL to sensing tasks, the paper aims to inspire further research in this domain. The discussion section has raised several important points for consideration. It highlighted the need to address the asynchrony in FL, leverage unlabelled on-device data, explore FL applications in Human Activity Recognition (HAR) and affective computing, and adapt FL methodologies from simulation environments to real-world applications on sensing devices. Efficient wireless communication, resource efficiency, and personalization are crucial when applying FL methodologies to sensing devices. Future research should focus on developing wireless communication techniques that minimize data exchange, designing algorithms with resource constraints in mind, and enabling privacy-preserving personalization. By addressing these challenges, FL methodologies can be effectively applied in real-world sensing applications, considering sensing devices’ unique constraints and requirements. Moreover, the discussion emphasizes the need for studies that bridge the gap between simulation and real-world deployment, ensuring the practicality and effectiveness of FL methodologies in real-world sensing applications. By adapting FL methodologies to real-world settings, researchers can address additional challenges, such as communication constraints, privacy concerns, and device heterogeneity, and validate the performance of FL in diverse sensing domains. In conclusion, this paper aims to provide a comprehensive review for researchers and practitioners interested in applying FL methodologies to edge sensing devices to contribute to developing intelligent and efficient systems in diverse application domains by highlighting the specific challenges and opportunities associated with these devices. Acknowledgment T¨ubitak Bideb 2211-A academic reward is gratefully acknowledged. This research has been supported by the Bo˘gazi¸ci University Research Fund, project number: 19301P. References S. Dey, A. Mukherjee, H. S. Paul, A. Pal, Challenges of using edge devices in iot computation grids, in: 2013 International Conference on Parallel and Distributed Systems, IEEE, 2013, pp. 564–569. L. Atzori, A. Iera, G. Morabito, The internet of things: A survey, Computer networks 54 (2010) 2787–2805. P. Voigt, A. Von dem Bussche, The eu general data protection regulation (gdpr), A Practical Guide, 1st Ed., Cham: Springer International Publishing 10 (2017) 10–5555. M. Capra, R. Peloso, G. Masera, M. Ruo Roch, M. Martina, Edge computing: A survey on the hardware requirements in the internet of things world, Future Internet 11 (2019) 100. Q. Yang, Y. Liu, T. Chen, Y. Tong, Federated machine learning: Concept and applications, ACM Transactions on Intelligent Systems and Technology (TIST) 10 (2019) 1–19. W. Yu, F. Liang, X. He, W. G. Hatcher, C. Lu, J. Lin, X. Yang, A survey on the edge computing for the internet of things, IEEE access 6 (2017) 6900–6919. D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, H. V. Poor, Federated learning for internet of things: A comprehensive survey, IEEE Communications Surveys & Tutorials 23 (2021) 1622–1658. M. Aledhari, R. Razzak, R. M. Parizi, F. Saeed, Federated learning: A survey on enabling technologies, protocols, and applications, IEEE Access 8 (2020) 140699–140725. J. Xu, B. S. Glicksberg, C. Su, P. Walker, J. Bian, F. Wang, Federated learning for healthcare informatics, Journal of Healthcare Informatics Research 5 (2021) 1–19. Q. Wu, K. He, X. Chen, Personalized federated learning for intelligent iot applications: A cloud-edge based framework, IEEE Open Journal of the Computer Society 1 (2020) 35–44. E. Sannara, F. Portet, P. Lalanda, V. German, A federated learning aggregation algorithm for pervasive computing: Evaluation and comparison, in: 2021 IEEE International Conference on Pervasive Computing and Communications (PerCom), IEEE, 2021, pp. 1–10. F. Wang, A. Preininger, Ai in health: state of the art, challenges, and future directions, Yearbook of medical informatics 28 (2019) 016–026. N. Rieke, J. Hancox, W. Li, F. Milletari, H. R. Roth, S. Albarqouni, S. Bakas, M. N. Galtier, B. A. Landman, K. Maier-Hein, et al., The future of digital health with federated learning, NPJ digital medicine 3 (2020) 119. M. Aledhari, R. Razzak, R. M. Parizi, F. Saeed, Federated learning: A survey on enabling technologies, protocols, and applications, IEEE Access 8 (2020) 140699–140725. Q. Jing, W. Wang, J. Zhang, H. Tian, K. Chen, Quantifying the performance of federated transfer learning, arXiv preprint arXiv:1912.12795 (2019). J. Yao, M. Mouhoub, A. Manashty, Y. Morgan, et al., Analysis of Model Aggregation Techniques in Federated Learning, Ph.D. thesis, Faculty of Graduate Studies and Research, University of Regina, 2021. P. Kairouz, H. B. McMahan, B. Avent, A. Bellet, M. Bennis, A. N. Bhagoji, K. Bonawitz, Z. Charles, G. Cormode, R. Cummings, et al., Advances and open problems in federated learning, Foundations and Trends® in Machine Learning 14 (2021) 1–210. A. Saeed, F. D. Salim, T. Ozcelebi, J. Lukkien, Federated self-supervised learning of multisensor representations for embedded intelligence, IEEE Internet of Things Journal 8 (2020) 1030–1040. T. Li, A. K. Sahu, A. Talwalkar, V. Smith, Federated learning: Challenges, methods, and future directions, IEEE signal processing magazine 37 (2020) 50–60. J. Koneˇcn`y, H. B. McMahan, F. X. Yu, P. Richt´arik, A. T. Suresh, D. Bacon, Federated learning: Strategies for improving communication efficiency, arXiv preprint arXiv:1610.05492 (2016). K. Wang, R. Mathews, C. Kiddon, H. Eichner, F. Beaufays, D. Ramage, Federated evaluation of on-device personalization, arXiv preprint arXiv:1910.10252 (2019). K. Bonawitz, H. Eichner, W. Grieskamp, D. Huba, A. Ingerman, V. Ivanov, C. Kiddon, J. Koneˇcn`y, S. Mazzocchi, B. McMahan, et al., Towards federated learning at scale: System design, Proceedings of machine learning and systems 1 (2019) 374–388. K. Bonawitz, V. Ivanov, B. Kreuter, A. Marcedone, H. B. McMahan, S. Patel, D. Ramage, A. Segal, K. Seth, Practical secure aggregation for privacypreserving machine learning, in: proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, 2017, pp. 1175– 1191. Z. Zhou, S. Yang, L. Pu, S. Yu, Cefl: Online admission control, data scheduling, and accuracy tuning for cost-efficient federated learning across edge nodes, IEEE Internet of Things Journal 7 (2020) 9341–9356. Y. Jiang, R. Cong, C. Shu, A. Yang, Z. Zhao, G. Min, Federated learning based mobile crowd sensing with unreliable user data, in: 2020 IEEE 22nd International Conference on High Performance Computing and Communications; IEEE 18th International Conference on Smart City; IEEE 6th International Conference on Data Science and Systems (HPCC/SmartCity/DSS), IEEE, 2020, pp. 320–327. B. McMahan, E. Moore, D. Ramage, S. Hampson, B. A. y Arcas, Communication-efficient learning of deep networks from decentralized data, in: Artificial intelligence and statistics, PMLR, 2017, pp. 1273–1282. N. Shlezinger, M. Chen, Y. C. Eldar, H. V. Poor, S. Cui, Uveqfed: Universal vector quantization for federated learning, IEEE Transactions on Signal Processing 69 (2020) 500–514. S. P. Karimireddy, S. Kale, M. Mohri, S. Reddi, S. Stich, A. T. Suresh, Scaffold: Stochastic controlled averaging for federated learning, in: International conference on machine learning, PMLR, 2020, pp. 5132–5143. A. Reisizadeh, A. Mokhtari, H. Hassani, A. Jadbabaie, R. Pedarsani, Fedpaq: A communication-efficient federated learning method with periodic averaging and quantization, in: International Conference on Artificial Intelligence and Statistics, PMLR, 2020, pp. 2021–2031. H. Wang, M. Yurochkin, Y. Sun, D. Papailiopoulos, Y. Khazaeni, Federated learning with matched averaging, arXiv preprint arXiv:2002.06440 (2020). M. Zaharia, A. Konwinski, A. D. Joseph, R. H. Katz, I. Stoica, Improving mapreduce performance in heterogeneous environments., in: Osdi, volume 8, 2008, p. 7. C. Xie, S. Koyejo, I. Gupta, Asynchronous federated optimization, arXiv preprint arXiv:1903.03934 (2019). W. Wu, L. He, W. Lin, R. Mao, C. Maple, S. Jarvis, Safa: A semiasynchronous protocol for fast federated learning with low overhead, IEEE Transactions on Computers 70 (2020) 655–668. M. Duan, D. Liu, X. Chen, Y. Tan, J. Ren, L. Qiao, L. Liang, Astraea: Selfbalancing federated learning for improving classification accuracy of mobile deep learning applications, in: 2019 IEEE 37th international conference on computer design (ICCD), IEEE, 2019, pp. 246–254. J. Liu, J. H. Wang, C. Rong, Y. Xu, T. Yu, J. Wang, Fedpa: An adaptively partial model aggregation strategy in federated learning, Computer Networks 199 (2021) 108468. Y. Gao, L. Liu, X. Zheng, C. Zhang, H. Ma, Federated sensing: Edgecloud elastic collaborative learning for intelligent sensing, IEEE Internet of Things Journal 8 (2021) 11100–11111. I. Erg¨un, H. U. Sami, B. G¨uler, Communication-efficient secure aggregation for federated learning, in: GLOBECOM 2022-2022 IEEE Global Communications Conference, IEEE, 2022, pp. 3881–3886. C. He, S. Li, J. So, X. Zeng, M. Zhang, H. Wang, X. Wang, P. Vepakomma, A. Singh, H. Qiu, et al., Fedml: A research library and benchmark for federated machine learning, arXiv preprint arXiv:2007.13518 (2020). J. Xu, S. Wang, L. Wang, A. C.-C. Yao, Fedcm: Federated learning with client-level momentum, arXiv preprint arXiv:2106.10874 (2021). T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, V. Smith, Federated optimization in heterogeneous networks, Proceedings of Machine learning and systems 2 (2020) 429–450. T. Semwal, A. Mulay, A. M. Agrawal, Fedperf: A practitioners’ guide to performance of federated learning algorithms (2020). I. Kholod, E. Yanaki, D. Fomichev, E. Shalugin, E. Novikova, E. Filippov, M. Nordlund, Open-source federated learning frameworks for iot: A comparative review and analysis, Sensors 21 (2020) 167. TFF, Tensorflow federated: Machine learning on decentralized data, Accessed: 21 September 2023. https://www.tensorflow.org/federated. S. Caldas, S. M. K. Duddu, P. Wu, T. Li, J. Koneˇcn`y, H. B. McMahan, V. Smith, A. Talwalkar, Leaf: A benchmark for federated settings, arXiv preprint arXiv:1812.01097 (2018). O. Community, Open-source python 3 based library developed by the openmined community, Accessed: 21 September 2023. https://blog. openmined.org/tag/pysyft/. D. J. Beutel, T. Topal, A. Mathur, X. Qiu, J. Fernandez-Marques, Y. Gao, L. Sani, K. H. Li, T. Parcollet, P. P. B. de Gusm˜ao, et al., Flower: A friendly federated learning research framework, arXiv preprint arXiv:2007.14390 (2020). Fedai, An industrial grade federated learning framework, Accessed: 21 September 2023. https://fate.fedai.org/. Intel, Open-source python 3 based library for federated learning developed by the intel, Accessed: 21 September 2023. https://github.com/intel/ openfl. M. Popovic, M. Popovic, I. Kastelan, M. Djukic, S. Ghilezan, A simple python testbed for federated learning algorithms, arXiv preprint arXiv:2305.20027 (2023). H. Cho, A. Mathur, F. Kawsar, Flame: Federated learning across multidevice environments, Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 6 (2022) 1–29. R. Jaren, E. Scott, L. Chapin, The internet of things: an overviewunderstanding the issues and challenges of a more connected world, Internet Soc (2015). R. Want, B. N. Schilit, S. Jenson, Enabling the internet of things, Computer 48 (2015) 28–35. H. U. Rehman, M. Asif, M. Ahmad, Future applications and research challenges of iot, in: 2017 International conference on information and communication technologies (ICICT), IEEE, 2017, pp. 68–74. J. McCann, L. Quinn, S. McGrath, E. O’Connell, Towards the distributed edge–an iot review, in: 2018 12th International Conference on Sensing Technology (ICST), IEEE, 2018, pp. 263–268. J. Feng, C. Rong, F. Sun, D. Guo, Y. Li, Pmf: A privacy-preserving human mobility prediction framework via federated learning, Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 4 (2020) 1–21. K. Sozinov, V. Vlassov, S. Girdzijauskas, Human activity recognition using federated learning, in: 2018 IEEE Intl Conf on Parallel & Distributed Processing with Applications, Ubiquitous Computing & Communications, Big Data & Cloud Computing, Social Computing & Networking, Sustainable Computing & Communications (ISPA/IUCC/BDCloud/SocialCom/SustainCom), IEEE, 2018, pp. 1103– 1111. S. Ek, F. Portet, P. Lalanda, G. Vega, Evaluation of federated learning aggregation algorithms: application to human activity recognition, in: Adjunct proceedings of the 2020 ACM international joint conference on pervasive and ubiquitous computing and proceedings of the 2020 ACM international symposium on wearable computers, 2020, pp. 638–643. C. Srinivasan, B. Rajesh, P. Saikalyan, K. Premsagar, E. S. Yadav, A review on the different types of internet of things (iot), Journal of Advanced Research in Dynamical and Control Systems 11 (2019) 154–158. D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, H. V. Poor, Federated learning for internet of things: A comprehensive survey, IEEE Communications Surveys & Tutorials 23 (2021) 1622–1658. G. Szegedi, P. Kiss, T. Horv´ath, Evolutionary federated learning on eegdata., in: ITAT, 2019, pp. 71–78. T. Wang, Y. Du, Y. Gong, K.-K. R. Choo, Y. Guo, Applications of federated learning in mobile health: Scoping review, Journal of Medical Internet Research 25 (2023) e43006. G. O. Ogbuabor, J. C. Augusto, R. Moseley, A. van Wyk, Context-aware support for cardiac health monitoring using federated machine learning, in: Artificial Intelligence XXXVIII: 41st SGAI International Conference on Artificial Intelligence, AI 2021, Cambridge, UK, December 14–16, 2021, Proceedings 41, Springer, 2021, pp. 267–281. R. W. Picard, Affective computing, MIT press, 2000 Y. S. Can, C. Ersoy, Privacy-preserving federated deep learning for wearable iot-based biomedical monitoring, ACM Transactions on Internet Technology (TOIT) 21 (2021) 1–17. D.-V. Nguyen, K. Zettsu, Spatially-distributed federated learning of convolutional recurrent neural networks for air pollution prediction, in: 2021 IEEE International Conference on Big Data (Big Data), IEEE, 2021, pp. 3601–3608. S. Chen, G. Long, T. Shen, J. Jiang, Prompt federated learning for weather forecasting: Toward foundation models on meteorological data, arXiv preprint arXiv:2301.09152 (2023). J. So, B. G¨uler, A. S. Avestimehr, Turbo-aggregate: Breaking the quadratic aggregation barrier in secure federated learning, IEEE Journal on Selected Areas in Information Theory 2 (2021) 479–489. J. C. Jiang, B. Kantarci, S. Oktug, T. Soyata, Federated learning in smart city sensing: Challenges and opportunities, Sensors 20 (2020) 6230. Y. Liu, J. James, J. Kang, D. Niyato, S. Zhang, Privacy-preserving traffic flow prediction: A federated learning approach, IEEE Internet of Things Journal 7 (2020) 7751–7763. D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, H. V. Poor, Federated learning for internet of things: A comprehensive survey, IEEE Communications Surveys & Tutorials 23 (2021) 1622–1658. A. M. Elbir, B. Soner, S. C¸ ¨oleri, D. G¨und¨uz, M. Bennis, Federated learning in vehicular networks, in: 2022 IEEE International Mediterranean Conference on Communications and Networking (MeditCom), IEEE, 2022, pp. 72–77. X. Liang, Y. Liu, T. Chen, M. Liu, Q. Yang, Federated transfer reinforcement learning for autonomous driving, in: Federated and Transfer Learning, Springer, 2022, pp. 357–371. Z. Ge, Z. Song, S. X. Ding, B. Huang, Data mining and analytics in the process industry: The role of machine learning, Ieee Access 5 (2017) 20590– 20616. V. Pruckovskaja, A. Weissenfeld, C. Heistracher, A. Graser, J. Kafka, P. Leputsch, D. Schall, J. Kemnitz, Federated learning for predictive maintenance and quality inspection in industrial applications, arXiv preprint arXiv:2304.11101 (2023). X. Han, H. Yu, H. Gu, Visual inspection with federated learning, in: Image Analysis and Recognition: 16th International Conference, ICIAR 2019, Waterloo, ON, Canada, August 27–29, 2019, Proceedings, Part II 16, Springer, 2019, pp. 52–64. F. Liu, X. Wu, S. Ge, W. Fan, Y. Zou, Federated learning for vision-andlanguage grounding problems, in: Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, 2020, pp. 11572–11579. S. Zhai, X. Jin, L. Wei, H. Luo, M. Cao, Dynamic federated learning for gmec with time-varying wireless link, IEEE Access 9 (2021) 10400–10412. N. I. Mowla, N. H. Tran, I. Doh, K. Chae, Federated learning-based cognitive detection of jamming attack in flying ad-hoc network, IEEE Access 8 (2019) 4338–4350. W. Zhou, Y. Li, S. Chen, B. Ding, Real-time data processing architecture for multi-robots based on differential federated learning, in: 2018 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computing, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI), IEEE, 2018, pp. 462–471. B. Liu, L. Wang, M. Liu, C.-Z. Xu, Federated imitation learning: A novel framework for cloud robotic systems with heterogeneous sensor data, IEEE Robotics and Automation Letters 5 (2020) 3509–3516. B. Liu, L. Wang, M. Liu, Lifelong federated reinforcement learning: a learning architecture for navigation in cloud robotic systems, IEEE Robotics and Automation Letters 4 (2019) 4555–4562. M. Fu, Y. Shi, Y. Zhou, Federated learning via unmanned aerial vehicle, IEEE Transactions on Wireless Communications (2023). W. Li, W. Yang, G. Jin, J. Chen, J. Li, R. Huang, Z. Chen, Clustering federated learning for bearing fault diagnosis in aerospace applications with a self-attention mechanism, Aerospace 9 (2022) 516. T. Zeng, O. Semiari, M. Mozaffari, M. Chen, W. Saad, M. Bennis, Federated learning in the sky: Joint power allocation and scheduling with uav swarms, in: ICC 2020-2020 IEEE International Conference on Communications (ICC), IEEE, 2020, pp. 1–6. J. Ebenezer, P. G. Krishna, M. Poojitha, A. V. Krishna, Plant leaf disease detection and classification with cnn and federated learning approach, in: Conference of Innovative Product Design and Intelligent Manufacturing System, Springer, 2022, pp. 513–523. A. Abu-Khadrah, A. M. Ali, M. Jarrah, An amendable multi-function control method using federated learning for smart sensors in agricultural production improvements, ACM Transactions on Sensor Networks (2023). This paper is under a CC 4.0 license. available on Arxiv

Stories, News & Mentions about JARVISML

Jarvisml’s Stories on HackerNoon