Mindtera

mindtera

Hello! My name is Amir, I am a social media & community manager, and I spend a lot of time in front of the computer: eight hours a day, sometimes more. Three months ago, to better understand what happens to my brain when I work, I bought the Waverox Mind Tracker and spent over 90 hours in it. In this review, I will tell you how it works, its pros and cons, and who it might be useful for. What is Mind Tracker and why did I buy it? Essentially, the Waverox Mind Tracker is a headband that helps track your brain’s state throughout the day. Waverox says that the gadget will help improve productivity, and teach you how to focus better and cope with stress. Honestly, I don't have serious productivity problems, and I don't generally stress too much, but I was curious about what happens in my head and whether I could work even more efficiently. That's how I got the device. Before this, I had never used anything like that, other than fitness trackers for sport. Price, form factor, design, and other specifications I bought MindTracker in October 2023, and it cost me ~700 Euros at the time. I don't know the current price – but you can learn it by leaving a request on the Waverox website. You can choose from 2 form factors: wireless headphones with noise cancellation, or a headband (my option). As far as I’m aware, their functionality is pretty much identical, and they have the same number of electrodes. I’ve picked the headband because it’s slightly lighter, and I wouldn’t trade my AirPods for the world. The gadget was delivered in a minimalistic matte black box, with a travel case inside. The first impression was rather positive: the headband is very good-looking, made of matte plastic, and the wires are hidden inside the headband. Golden electrodes that do most of the work are also hidden and aren’t seen when I wear the headband. I could see that engineers and designers did a great job. Ergonomics-wise, the headband is very light and small, so I never had any problems with wearing it outside. Additionally, you can adjust the headband to your head. A USB charger is included in the packaging. I charge the headband about once a week, and I wear it for 1-2 hours on weekdays, so I just charge it on Sunday. To use the Mind Tracker, you can either use a PC or a smartphone. I use both the Mac OS and the IOS apps, and it synchronizes great. I always have the app open when I’m working, but outside of work, for a meditation session or to check the data for the day, I use the mobile app. What MindTracker does and how I use it The device reads real-time brain activity data using electrodes. It processes the obtained data and translates it into metrics: concentration, fatigue, stress level, cognitive load level, and others. Right from the home screen, I can see three blocks of data about my state. The first one shows the brain's load level (usually represented as a percentage). This metric shows me how busy my brain is at the moment. One funny experiment I did is trying to imagine a complex environment while closing my eyes - my brain’s load level skyrockets. The second block reflects my overall current state: engagement, relaxation, anxiety, stress, or overwork. In three months, it showed a state of fatigue only once; the rest of the time, I was either focused or relaxed. Based on this metric, you can make more informed decisions about when to work. Usually, when I see that my brain is focused, I try to get as much work done as possible. The third block shows my emotional state: whether I am focused, calm, or tense. I found this metric especially useful, as I tend not to give attention to my emotions when heavily concentrated. If the app shows me I’m tense, it’s a sign for me to take a step back, and think about why I feel that way. To ensure accurate tracking, I calibrate my device in the morning: I put it on while being in a calm state and just measure the indicators. This allows me to track iAPF: my If it decreases, something is wrong, and I need to rest. It's interesting to measure iAPF in the morning and evening; for example, this morning, my iAPF was -3.6%, and now, in the evening, it's -2%. individual alpha frequency. I also go through the Stress Resistance training. It is a 25-day course of nine-minute meditations. I usually do it like this: dim the lights, go to a quiet place, sit in one position, and listen to the training. During the session, MindTracker reads the brain's state and plays different sounds. If I am completely relaxed, it suggests nature sounds; if I stop feeling relaxed, it adds the sounds of a busy highway. Ideally, the training should be done every day, but I don't always find the time, so I have only reached level 6 so far. At the end of the session, the device informs whether I passed the level. If the training is not done for some time, the statistics are reset, and you have to start over. Here’s how I use the Mind Tracker. I wake up in the morning, calibrate it, then work for 3-4 hours. It’s important to note that I work from home, so I calibrate the device BEFORE starting to work. I usually leave it on when I do my usual coffee run since it’s interesting to me how my brain switches once I’m out of the house. I have no problems with wearing the headband outside: it’s not at all heavy and feels like it’s made out of fabric. Towards the evening, I sometimes put it back on to do the training I mentioned earlier. When working from a coffee shop or a co-working, the headband doesn't bother anyone. Sometimes people ask what it is because the device has a pretty noticeable green light when turned on. Personally, I like the way the device looks, so I like wearing it in public. Seems very cyberpunk-ish, especially with a black hoodie and a laptop. In conclusion, the Waverox Mind Tracker didn’t make any shocking revelations about me, or the way I work. I’m a pretty calm and productive person, and the device mainly confirmed that for me. However, I find it very useful for deciding when to work and when to take breaks. I’ve tried the Pomodoro Technique before using the Mind Tracker, but found it lacking personalization. For this purpose, the Waverox Mind Tracker is perfect for me. Pros, cons, and who it's suitable for Mind Tracker pros: Excellent design. Accurately shows the state. Various metrics. Training programs: perfect solution for those who want to start meditating, learn to relax, and deal with stress. Cons: Form factor: it would be great if the device was even smaller and could fit into the ear. No app for Android yet. Calibration is needed, and while it takes a minute to perform, I’m still too lazy for that sometimes.

That One Time I Tried To Hack My Brain

সেই এক সময় আমি আমার মস্তিষ্ক হ্যাক করার চেষ্টা করেছি

Das eine Mal, als ich versuchte, mein Gehirn zu hacken

Esa vez traté de hackear mi cerebro

Cette fois-là, j'ai essayé de pirater mon cerveau

उस एक बार मैंने अपना दिमाग हैक करने की कोशिश की थी

ある時、私は自分の脳をハッキングしようとした

그 때 나는 내 두뇌를 해킹하려고 시도했습니다.

Aquela vez que tentei hackear meu cérebro

Тот раз, когда я пытался взломать свой мозг

Bir Kez Beynimi Hacklemeye Çalışmıştım

Đó Một Lần Tôi Đã Cố Hack Bộ Não Của Mình

那一次我试图破解我的大脑

DeFi, CeFi, and now ReFi – . But, behind these acronyms are important trends that are currently reshaping the way we use and move money, the way we invest, where capital is allocated, and to what end. yes, web3 can be confusing sometimes ReFi, or Regenerative Finance, is a mission-driven movement standing at the frontier between crypto, impact investing, and sustainable business practices like . In short, ReFi is a movement leveraging blockchain to create financial incentives to lower carbon emissions, regenerate the environment and ultimately reverse climate change. ESG Yes, that’s ambitious. But, this fast-growing sector is already developing some of the key infrastructure and applications needed to make a regenerative economy significantly more attractive and drive positive climate action. Under the ReFi banner is a community of financial activists who are working to incentivize a healthy and equitable environmental system. The goal: redefining users’ relationship with finance and wealth. Even though ReFi is still an emerging movement today, it already boasts a thriving ecosystem of decentralized protocols, startups, non-profits, investors, VCs, and users – with many success stories. Why do we need regenerative finance? The crypto bear market has brought many questions with it. One of the biggest ones is: “What are the real-world use cases of web3?” Well, the blockchain ecosystem has shown time and time again that it is very effective at creating mechanisms that align economic incentives toward a common goal. ReFi simply provides a new common goal and leverages web3 mechanisms to reach it. So why is it important? The “regenerative” part refers to a self-renewing process that any system needs to thrive for long periods of time. It also refers to the ability to adapt to unexpected, sometimes threatening circumstances. No system can sustain itself over the long term if it is not designed to continuously regenerate. In contrast, today’s economy is mainly exploitative and extractive. This means that it treats most economies as places from which wealth (resources, money, labor, etc.) can be extracted. Extractive activities are common because they help build strength and power by concentrating resources. Applied to the world of Blockchain, these concepts highlight two things: First, a significant part of the crypto universe is extractive. Unsustainable tokenomics, energy waste, hype and speculation, lack of utility… From a cynical point of view, many web3 projects exist solely to extract wealth from unsuspecting investors. An obvious example is the ever-growing energy consumption of PoW mining. PoW goes against the principle of regeneration because it incentivizes the use of the cheapest energy in the largest quantities. Second, the good news is, it is possible to design blockchain systems that are human-centered, holistic, and designed for the long term. The incredible advantage of web3 lies in how it can incentivize regenerative behavior. Smart contracts, NFTs, tokens, and more can be used to drive our economy toward regeneration. But every organization has a role to play. Offsetting carbon emissions A big part of being regenerative is making sure you keep your carbon footprint in check. For ReFi companies, it’s usually not enough to be carbon neutral – they aim to be “carbon positive” or “climate positive”. Today, many organizations use different terms to describe how they tackle climate change. These labels are meant to offer an easy way to recognize their efforts to be sustainable. To be climate-positive means that a business removes more greenhouse gases than it produces. There are several ways to accomplish this: through significant carbon removals or avoided emissions. In crypto, avoiding emissions means opting for energy-efficient alternatives when available. And the trend is clearly in favor of “clean” technologies, as shown with the recent Ethereum Merge which reduced the network’s energy consumption by 99.95% by switching to Proof-of-Stake. Some companies make strong commitments to offset their emissions. For instance, at we pledge to donate 50% of the company’s profits to partner environmental organizations with a real positive impact on the planet. Mintera Incentivizing net positive actions With the UN Sustainable Development Goals (SDGs) set to guide global sustainability efforts over the next decade, it’s important to consider how different technologies can support these objectives. We know blockchain can be used to track environmental impact, enable transparency and accountability, and drive sustainable behavior change. Here are some examples of how blockchain can help create a more sustainable future: protocols can reward participants who are taking real action to remove CO2 emissions with tokens or NFTs with monetary value. Gamification: blockchains can track carbon emissions, and climate metrics to help hold governments, corporations, and NGOs accountable to climate goals. Accountability: by tokenizing carbon-sequestering assets such as rainforests, or carbon credits, we can incentivize more climate-friendly behaviors, while protecting natural assets. Tokenization: DAOs and other web3 organized communities provide a structure that can facilitate raising funds for climate causes and advocating for sustainable policies. Leveraging communities: And there are many more creative ways to use web3 for good. At we are launching a 5-years Grant Program to help kick-start and scale green projects initiated by our community. People who want to be rewarded for their climate actions can find many opportunities in the ReFi space. Mintera How you can use crypto in a sustainable way Contrary to some beliefs, it is entirely possible to invest, use and participate in crypto in ways that are beneficial to the environment. Here are a that you can act on right now: few tips Only use energy-efficient protocols: there are many options - in general, you want to avoid cryptocurrencies that are mined using power-hungry hardware. Explore the ReFi space to get a feel of the main projects and possibilities. You can start by having a look at the #ReFi hashtag on Twitter. Start offsetting your carbon emission using tokenized carbon credits with projects such as Toucan Protocol. Find a like-minded community by joining a DAO dedicated to the climate, like KlimaDAO or ReFi DAO. And remember, we are still early in the ReFi space. Don’t be afraid to try and learn, be creative, and even start your own projects!

Greed To Green: Inside The New Trend Of Regenerative Finance

If you live or work in Montreal, you've probably passed by one of the 700 Bixi stations. And you've undoubtedly come across "Bixists," i.e., cyclists riding one of the grey bikes (or perhaps a blue bike). That's Bixi. It is a hybrid between " cycle" and "Ta " to underline the concept of using a bicycle just like a taxi. BI XI While it has been growing in popularity, it would be an excellent opportunity to analyze various aspects of Bixi data, especially in Montreal. The goal is to find results that can help improve the quality of the service, help customers find their closest Bixi station, the most and least crowded time to pick up a Bixi, and all other services that will benefit the customer. For this particular case study, we will use open datasets. Datasets The datasets we will be using in this analysis are: OD_2016.csv OD_2017.csv Stations_2016.csv Stations_2017.csv noaa-daily-weather-data_2016.csv noaa-daily-weather-data_2017.csv List of Holidays The datasets regarding bixi ridership have over four million records. The data includes the start and end station code, the start and end date plus time, the duration in seconds, and membership details. The file with all the station details (Stations.csv) contains the station's details against their station codes like code, name, address, coordinates, active status. The weather files (noaa-daily-weather-data.csv) contain all-weather information throughout the year like date, precipitation, snow, max and min temperature, elevation, coordinates, and country code. We have the list of holidays and their corresponding date, as per the year, to know what the traffic is like on those days. Data Process After procuring all the data, we will follow these steps to complete our analysis. Here's what we will be finding out: Find the most popular and least popular bike stations. Find the most and least popular days of the week in terms of ridership. Find the most and least popular times of the day in terms of ridership. Find the most and least popular ridership by weather conditions. Which stations exhaust their bike capacity the fastest and when? Which stations have the most unused bikes? Are there any holidays apart from the weekends when bike ridership is popular Does it make a difference if you are a member or not Most and Least Popular Stations The first task is to find out the most and least popular bike stations. Popularity is measured by the number of bikes taken and returned at a station. For further simplicity, we will find four stations; Most Popular Station to Start - marked in green Least Popular Station to Start- marked in red Most Popular Station to End- marked in blue Least Popular Station to End - marked in yellow To complete this task, we will be getting all the data specifically for 2016 and 2017. Next, we will sort stations according to their popularity. We will calculate popularity by a simple method, the more times a station has a bike rented or returned, the more popular that particular station is. Following this step, we will find the most and least popular stations according to starting point and ending point. They are listed below, along with their name and code. For 2016 Most Popular Station to Start Mackay /de Maisonneuve (Sud) 6100 Least Popular Station to Start Place Longueuil 5003 Most Popular Station to End de Maisonneuve / de Bleury 6078 Least Popular Station to End Place Longueuil 5003 For 2017 Most Popular Station to Start Mackay /de Maisonneuve (Sud) 6100 Least Popular Station to Start Place Longueuil 5003 Most Popular Station to End Berri / de Maisonneuve 6015 Least Popular Station to End Place Longueuil 5003 Note: As the Station with the least popularity to start and end was the same, a single point represents it on the map. Most and Least Popular Days Now that we have the most and least popular stations in Montreal, we will next determine the most and the least popular 'Days of the Week' when the ridership was most popular. For this analysis, we will use data from the sources mentioned above (dataset OD.csv) for 2016 and 2017. As for how we will analyze- we will check every entry, aka every date when a bixi was rented. We will then sort the dates into days and note the number of rentals made on that specific day. As a result, we will know the most popular day and the number of rented bikes on that day. For 2016 Wednesday 632176 Rents Tuesday 627808 Rents Friday 613929 Rents Thursday 607976 Rents Monday 560265 Rents Saturday 503056 Rents Sunday 454870 Rents For 2017 Wednesday 636580 Rents Thursday 635897 Rents Tuesday 595485 Rents Saturday 570472 Rents Friday 553359 Rents Sunday 528526 Rents Monday 498403 Rents What do these numbers depict? It shows that maximum rentals are made on Wednesday, with a whopping number of 636580 in 2016 and 632176 in 2017.  Surprisingly, the days that see the least number of rentals are on Sundays in 2016. However, Mondays see the least number of rentals in 2017, making sense as it is the beginning of a workweek. Most and Least Popular Times of Day We now know the most and least popular stations and the days of the week for ridership. The following agenda is to find the most and least popular times of day when a bike is rented. For this analysis, we will be using the above dataset (OD.csv), which gives us an hourly timestamp and also looks at 24 hours throughout the year. This analysis aims to help us get to the most popular hour when rents were made and the time with least rents for both 2016 and 2017. For 2016 Most Rents at 17:00 - 17:59 (425975 Rents) Least Rents at 04:00 - 04:59 (11340 Rents) For 2017 Most Rents at 17:00 - 17:59 (426821 Rents) Least Rents at 04:00 - 04:59 (11478 Rents) After looking at the analysis, it is clear that most rents were made at 5:00 PM throughout the year, which is the perfect time to go for a ride. As for the least rents, 4:00 AM was when bixis were least rented, which is not surprising. This analysis clearly shows that the evenings are when the stations are crowded. Weather Impact on Ridership Other than factors, one of the primary parameters is weather that can affect ridership. For this particular case - we will use two datasets - one with weather details (noaa-daily-weather-data.csv) and the other which has ridership details (OD.csv) for respective years. We will audit and compare them both to find any visible differences in ridership as weather changes. We first see the number of rides taken on a particular day and their duration, then check the weather conditions for that day and plot them against a graph to view any visible changes. We display the results below; 2016  and 2017: We have four parameters shown in the graphs (for 2016 and 2017) - ridership popularity, precipitation analysis, temperature analysis, and snow analysis. When you compare all the parameters, the results are as follows: The popularity of ridership is high during higher temperatures - we are talking sunny and spring days. Certain days see the effects of precipitation, but if you love taking a ride while it showers, this time is perfect for you. The number of rides drops when it snows. Stations that Exhaust Fastest and Slowest We don't want a day when you head to the station and don't find a ride. One would assume that the most popular stations would exhaust the fastest, and those that are least popular would finish the slowest. While this is true in most cases, the popularity analysis is annual, and our bixi case study results may vary. To solve this puzzle, we will first check the number of rentals and returns in a day at a particular station, then compare them with their maximum capacity. If the rentals exceed the maximum capacity, it is safe to say that the station exhausts their bikes the fastest. If it is the other way round, then the station finishes its bikes the slowest. Using the above data, the analysis leads us to the following: For 2016 du Mont-Royal / Vincent-d'Indy 6306 Exhaust Rate 2.46 Villa Maria (DÃ©carie / Monkland) 6101 Exhaust Rate 2.40 Louis-Colin / McKenna 6928 Exhaust Rate 2.17 Parc Jean-Brillant (Swail / Decelles) 6316 Exhaust Rate 2.12 de Darlington 6310 Exhaust Rate 1.98 … Notre-Dame / Peel 6085 Exhaust Rate 0.58 St-Denis / de Maisonneuve 6014 Exhaust Rate 0.53 Viger/Jeanne Mance 6035 Exhaust Rate 0.52 Place-d'Armes (Viger / St-Urbain) 6032 Exhaust Rate 0.51 Square Victoria 6043 Exhaust Rate 0.44 For 2017 Villa Maria (DÃ©carie / de Monkland) 6101 Exhaust Rate 2.54 Jean-Brillant / McKenna 6928 Exhaust Rate 2.43 du Mont-Royal / Vincent-d'Indy 6306 Exhaust Rate 2.31 de Darlington 6310 Exhaust Rate 1.97 Gatineau / Swail 6316 Exhaust Rate 1.97 … Lucien L'Allier / St-Jacques 6096 Exhaust Rate 0.58 St-Urbain 6034 Exhaust Rate 0.55 Notre-Dame / Peel 6085 Exhaust Rate 0.55 Place-d'Armes (Viger / St-Urbain) 6032 Exhaust Rate 0.48 Square Victoria 6043 Exhaust Rate 0.45 If a station's exhaustion rate is 2, the number of rents is two times faster than the number of returns, which means that more bikes are taken than returned, and that station is likely to exhaust its bikes fast. If a station's exhaust rate is 0.5, the number of returns is twice the number of rents, which means that more bikes are returned than rented, and that station will have more bikes than its capacity causing it never to exhaust. Holidays Other than Weekends with Popular Ridership For this particular analysis, we will be using the holiday data from 2016 and 2017. But we will only be using the holidays mentioned . Firstly, we will check if there are weekends and get rid of them. Second, we will check their dates, compare them with the primary data mentioned above for the respective years, and finally find the popularity of overall ridership on that day. here Once these steps are completed, we will compare the holidays with the primary data, which will determine the most popular ridership, and which has the least. We will look at holidays such as Victoria Day, St. Jean Baptiste Day, Canada Day, Labour Day, and Thanksgiving Day. Here are the corresponding dates. For 2016 St. Jean Baptiste Day 20095 Rents Victoria Day 19109 Rents Labour Day 18534 Rents Canada Day 17083 Rents Thanksgiving Day 9640 Rents For 2017 St. Jean Baptiste Day 25096 Rents Canada Day 16162 Rents Labour Day 14802 Rents Easter Monday 8036 Rents Victoria Day 7961 Rents Members and Non-Members in Ridership Does being a member or a non-member have any influence on the analysis thus far? Let's find out. We again use one of the data sets mentioned in the Data Source section. We find the member's and non-members' data and compare it to the number of rents. We calculated and compared the total number of rents and popularity with members and non-members, and the results are as follows for both years. 83% members 17%, not members In conclusion, 83% of the members played an essential part in renting the bikes and impacted the overall bixi traffic analysis. Top 4 interesting observations Surprisingly, the most popular time for bike rentals was not in the morning but in the evening. We expect that the morning rush hour to be the busiest, and our hypothesis was incorrect. We expected that Mondays might be the most popular day with people struggling to get into the offices on time after the weekend. Our hypothesis was wrong as Wednesdays were the most popular days of the week. We expected to see significant potential for usage on statutory holidays and weekends. Our hypothesis was true. As most of the use is member-driven (835), we think there is a significant opportunity to encourage BIXI usage for tourists and one-off usage by folks coming into the city for weekend activities, especially during the summer. We did analyze stations based on how quickly they exhaust bikes. We cannot tell if this results in mixed ridership opportunities and if there's a response mechanism to move the bikes back to a transit stop - which sees riders but no bikes and if such a mechanism responds to unmet demand. If you are from Montreal and have additional local insight, we look forward to hearing from you in the comments below. Code: https://github.com/Mindtrades-Consulting/Analyzing-BIXI-Ridership-With-Data How can MindTrades help? This case study is only a tipping point to such in-depth analysis with insight and solutions. For more such case studies, contact . https://www.mindtrades.com

Analyzing Montreal’s BIXI Ridership With Data And Visuals 

Scientific American, Volume XXIV., No. 12, March 18, 1871 by Various, is part of the HackerNoon Books Series. You can jump to any chapter in this book . THE MINERAL RESOURCES OF MISSOURI. here THE MINERAL RESOURCES OF MISSOURI. The incalculable wealth, which lies hid in the bosom of Mother Earth, in our vast possessions of the West, is undoubtedly centered in the State of Missouri; and the development of this fund of riches must add to the national prosperity, not only by its immeasurable intrinsic value, but by its affording occupation to armies of laborers, the latter being the highest and most important consideration. In 1852-3, a geological survey of the State was wisely decided upon, and a liberal provision for its execution made. Two valuable reports, by Professor Swallow, have been printed, in the year 1855, but the notes of his subsequent investigations have not been made public. In the session of 1869-70, further action, in this important public work, was taken by the State legislature, and arrangements made for a still more accurate and detailed examination, under the direction of Professor A. D. Hager, of Vermont. The distribution of metals all over the State will be seen in the following figures, taken from the St. Louis , which show the number of counties in which the various ores are found: Iron in 46 counties, lead in 43, coal in 36, copper in 24, marble in 11, zinc in 27, fire clay in 16, barytes in 10, nickel in 6, granite in 4, tin in 4, plumbago in 2, gypsum in 2, alum in 1, antimony in 4. Journal of Commerce There is probably no country in the world so endowed as this. Of iron alone, according to the State geologist's report for 1855, there is ore of the best quality, sufficient to furnish 200,000,000 tuns of iron; and this quantity lies in a small space, in the vicinity of Pilot Knob and Iron Mountain, and within 100 miles of St. Louis. The quality of the iron is highly spoken of by the manufacturers, and the capacity of the smelting appliances has reached to over 150,000 tuns per annum. The coal is well suited for reduction of ores, either by hot or cold blast treatment. The Scotia Iron Co. commenced operations in January, 1870; and, although the materials for building blast furnaces had to be carried 80 miles into a desert, the first furnace was blown into blast in August, 1870. This furnace will run about 24 tuns per day. The company procures ore from a hill, near the furnace, in which there is an apparently inexhaustible supply of red oxide and brown specular. This ore yields 60 per cent of pure metal. The erection of mills for making wrought iron is contemplated, and the high quality and prodigious quantity of the raw material will justify and reward any outlay of capital in this direction. The shipment of ore to other States goes on constantly, the last year's account showing that 246,555 tuns were dispersed over Indiana, Ohio, and others. The furnaces at Kingsland, South St. Louis, Lewis Iron Co.'s Works, Carondelet, and Maramec are all well situated as to coal and limestone, the Maramec Works having a most valuable water-power. These latter works also ship about 40,000 tuns red hematite ore yearly. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Various (2006). Scientific American, Volume XXIV., No. 12, March 18, 1871. Urbana, Illinois: Project Gutenberg. Retrieved https://www.gutenberg.org/cache/epub/19180/pg19180-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at . www.gutenberg.org https://www.gutenberg.org/policy/license.html

THE MINERAL RESOURCES OF MISSOURI

This newest podcast episode features Felix Tao, CEO of Mindverse AI, and his years of experience working as a researcher at and Alibaba, mostly involved in language applications and AI. In this interview, Felix gives his insights on the evolution of AI, large language models, and finding the right balance between research and application. Facebook Here's a quick overview of the first few minutes of the interview before you commit to an insightful hour-long episode! First, Felix reflects on the changing landscape of research during his time at Facebook, where he worked on content understanding tools, and how the field has now advanced with foundation models like GPT3 shaping the way for AGI (Artificial General Intelligence). AI He answers questions such as: is a research-focused career valuable today? Felix says it depends on the area - foundational work and addressing large language model challenges remain critical, but job seekers might profit from gaining industry experience. We discuss his time at Alibaba, and he shares his goals of "waking up the consciousness" in machines by combining large language models with human-like frameworks - including memory, perception, and motor control. Exciting times! Felix highlights the back-and-forth pattern of AI development, moving between fragmented specialized tasks and unified general approaches, and believes the future may witness domain-specific AI for higher quality and better problem-solving abilities. Check out Felix Tao's thought-provoking journey in AI research and development, and discover how MindverseAI's MindOS is aiming to employ specialized AI for domain-specific tasks, in this captivating interview. Watch the Video https://youtu.be/6w7UHzS00lI?embedable=true&transcript=true You can also listen on or Spotify Apple podcasts

From Facebook to MindverseAI: Felix Tao's Insights on AI Evolution and the Future of Large Language 

ফেসবুক থেকে মাইন্ডভার্সএআই: এআই বিবর্তন এবং বৃহৎ ভাষার ভবিষ্যত সম্পর্কে ফেলিক্স টাও-এর অন্তর্দৃষ্টি

Von Facebook zu MindverseAI: Felix Taos Einblicke in die KI-Evolution und die Zukunft großer Sprachen

De Facebook a MindverseAI: las ideas de Felix Tao sobre la evolución de la IA y el futuro de los idiomas grandes

De Facebook à MindverseAI : les idées de Felix Tao sur l'évolution de l'IA et l'avenir du grand langage

फ़ेसबुक से माइंडवर्सएआई तक: एआई इवोल्यूशन और बड़ी भाषा के भविष्य पर फ़ेलिक्स ताओ की अंतर्दृष्टि

Facebook から MindverseAI まで: AI の進化と大規模言語の将来に関するフェリックス タオの洞察

Facebook에서 MindverseAI까지: AI 진화와 대규모 언어의 미래에 대한 Felix Tao의 통찰

Do Facebook ao MindverseAI: insights de Felix Tao sobre a evolução da IA e o futuro da linguagem grande

Facebook'tan MindverseAI'ye: Felix Tao'nun Yapay Zekanın Evrimi ve Büyük Dilin Geleceği Konusundaki Görüşleri

Từ Facebook đến MindverseAI: Những hiểu biết sâu sắc của Felix Tao về sự tiến hóa của AI và tương lai của ngôn ngữ lớn

从 Facebook 到 MindverseAI：Felix Tao 对 AI 进化和大语言未来的见解

Artificial Intelligence helped give Democrats a fundraising edge this election season, allowing the blue party to retain control of the Senate and blunt a Republican takeover of the House. While there were many other factors at play, the Democrats relied more heavily on AI in finding donors, according to professional fundraisers, bringing in more money from individual, small-dollar donors than their Republican rivals. “AI will play an increasingly important role in democracy going forward,” said Martin Kurucz, who runs , a Democratic data technology company in the fundraising space. Sterling Data Company Data analytics have been used in political fundraising for more than a decade, and bots – automated systems that post on social media - have long spread political messages on a large scale. But using AI to identify donors and maximize outreach efforts is relatively new. On the left side of the aisle, fundraisers are using artificial intelligence to predict who's likely to donate to their cause, while on the Republican side, fundraisers are using artificial intelligence to maximize the efficiency of outreach to established lists of donors. While the Republican approach may raise more dollars per donor, they have fewer donors to raise from. Consider John Fetterman, the Democrat who won a senate seat in Pennsylvania against Republican Mehmet Oz. , Fetterman raised more than $55 million in 2022 compared with by Mehmet Oz. More than half of Fetterman’s haul came from individual donors giving $200 or less, while only about a third of Oz’s contributions were from such small-dollar donors. According to the Federal Election Commission just over $15 million Mr. Fetterman, whose campaign used AI, was not alone. Most statewide Democrats outraised their GOP counterparts by large multiples in the small-dollar space. from the 2022 election cycle show that Democratic donor organizations such as the Democratic Congressional Campaign Committee, or DCCC, spent around $8.4 million on digital acquisition, while raising $85.6 million from unitemized small dollar contributions – about ten times what they spent. At the same time, Republican organizations such as the National Republican Congressional Committee spent around $32 million on digital acquisition and raised less than double that amount in unitemized small dollar contributions. Federal Elections Commission data While small-dollar donors on the left are typically more affluent than their Republican counterparts, the Democrat advantage is more than just demographics. For most of the past decade, digital ads have been the prevalent means to find donors. But fundraising has shifted to email and mobile acquisition as digital advertising returns withered following that limited the tracking capabilities of digital advertisers. A few firms soon realized that AI-driven donor targeting was the most effective and fastest means of email and mobile acquisition. Apple’s 2021 privacy measures “No data analyst in the world can sort through tens of thousands of potential donors and figure out which is most likely to give money,” said Kurucz, whose firm helped both the DCCC and Fetterman, “but AI can do this.” “We have reached a point where anyone who can use a spreadsheet can use machine learning for data-driven decision-making,” said , cofounder of Akkio. Jonathon Reilly AI models can spot larger trends, predict online donations, and recalibrate targeting in seconds, allowing campaigns to tailor their outreach in near real-time. “We generate more revenue per donor than Democrats across the board,” said Tom Newhouse, vice president of digital marketing at , but added that Apple’s iOS changes have hurt Republicans’ donor prospecting strategies. Convergence Media Sterling Data uses a no-code machine-learning website, , which allows users to drag and drop their data spreadsheets into the site. On the backend, Akkio analyzes the data and gives the user options for what they want to predict. In Sterling Data’s case, that is each person’s likelihood of giving money. Akkio.com Sterling Data has built up a database of about 30 million donors with 500 or more columns of information on everything from the kind of car each donor drives to which Netflix shows they watch. It runs a subset of the data through an Akkio model trained to spot likely donors for a particular candidate. The result is a prospective donor list ranked from most likely to least likely to give, allowing Sterling Data to focus their efforts without wasting time and money on the wrong people. “The ease of using AI-ranked lists means data firms selling donor prospects curated by AI can offer pricing that undercuts the rest of the market,” said Jacob Geers, who until recently served as managing director of Veracity Media, a digital strategy agency. AI tools like Akkio now allow organizations to scale in ways that were once impossible, crunching millions of data points in seconds and prioritizing the actions to take for maximum effect. “We have reached a point where anyone who can use a spreadsheet can use machine learning for data-driven decision-making,” said , cofounder of Akkio. “Our platform enables non-technical users to extract value from their data.” Jonathon Reilly Sterling Data then gives the prioritized contact list to the candidate's team, who contact donors in the most efficient way possible. “The whole point is predicting who will actually donate if my candidate reaches out,” Kurucz said, adding that Akkio’s model builds prospective donor lists that consistently raise twice as much as lists built with other methods. Over the last cycle, AI-driven donor acquisition provided by vendors like Sterling Data has changed industry wide expectations for the speed of a return on investment, said Kevin Massey, a partner in , a digital fundraising agency. In past cycles, he said, it was standard to expect around 6 months or more to go by before an email acquisition paid off in full for a campaign. “But with these new AI-driven acquisitions, we've seen that increase rapidly to one to three months,” Massey said. Momentum Campaigns The faster payoff allows campaigns to react more quickly and be smarter with their acquisition budgets because they're seeing results faster. Sterling Data spends thousands of hours calling potential high-dollar donors. The firm uses another form of AI called natural language processing to read call notes and glean new details about those donors. Emails are reserved for potential small-dollar donors. Sterling Data works with Democrats up and down the ballot, from congressional to city council candidates. “Targeting at scale is really hard and getting a hold of the donor, whether that's through an email or a phone number, is even harder,” Kurucz said. There are many different data sources from which to aggregate information on potential donors. “The more information you get on potential donors, the better the model is,” said Kurucz. Kurucz believes the no-code approach will spread because it puts the power in the hands of a daily practitioner who no longer needs expensive data scientists to achieve results. Kurucz noted that he can create and run his AI models using Akkio on a plane on his laptop. Convergence Media’s Newhouse says they use AI features in an email and SMS platform called to optimize the timing or mode of outreach, whether by email or text message, for example, which leads to higher donations per donor. But he said Republicans have yet to coalesce around a central set of tech tools that are shared across the board. Iterable “To broaden the Republican donor base, Republicans need to embrace artificial intelligence or and data modeling to identify new donors,” he said.

How Artificial Intelligence Swayed The Midterm Elections - And Will Become A Permanent Feature Of Democracy

Cómo la inteligencia artificial influyó en las elecciones intermedias y se convertirá en una característica permanente de la democracia

Comment l'intelligence artificielle a influencé les élections de mi-mandat - et deviendra une caractéristique permanente de la démocratie

आर्टिफिशियल इंटेलिजेंस ने मध्यावधि चुनावों को कैसे प्रभावित किया - और यह लोकतंत्र की एक स्थायी विशेषता बन जाएगा

人工知能が中間選挙をどのように動かしたか - そして民主主義の恒久的な特徴になる

Como a inteligência artificial influenciou as eleições de meio de mandato - e se tornará uma característica permanente da democracia

Trí tuệ nhân tạo đã ảnh hưởng như thế nào đến cuộc bầu cử giữa nhiệm kỳ - và sẽ trở thành một đặc điểm lâu dài của nền dân chủ

人工智能如何影响中期选举——并将成为民主的永久特征

DAOs, NFTs and the future of Web3.0 DAOs are a new form of organization that operates on the blockchain. They are autonomous and run without any human supervision or control. DAOs can be set up to run themselves in a pre-programmed way, with rules embedded in the code, which cannot be broken by humans or by outside interference. . They can be bought, sold, or traded with others. NFTs are stored in a decentralized ledger, which is accessible to anyone who has access to the internet. NFTs are a type of digital asset that exists on the blockchain Both DAOs and NFTs are amongst the hottest topics on the space of blockchain right now. So, we thought we’d interview Paul Dowling to learn more about this subject. Our Guest: Paul Dowling is the Founder and CEO of . Paul has over 30 years experience of working in technology and management consulting firms. He has also spent the last 10 years founders and investors in the European technology startup ecosystem. Paul Dowling Mindstream AI working with startup He is a regular speaker at international startup and mentors several leading accelerators. Paul is passionate about the potential for large and medium-sized organizations to innovate through artificial intelligence and insights gained through . events data science Paul, what does Mindstream do? PD:| Mindstream AI is a . Our Data-driven startups that want to grow and also help large corporations that might want to access some of the innovation which is coming from those startups. And then finally, we have an arm of our company that is helping people upscale with . startup and it is a startup that is involved in Data role is mainly to work with other companies in the ecosystem and try to help Data skills Do you know anything about DAOs and NTFs? PD:| Yes! So I have a second company that is called Dreamblocks and Dreamblocks is a Web 3.0 company that has the future intention of structuring itself as a DAO. DAO stands for Decentralised Autonomous Organisation and to create those companies is quite complex. You don’t just register a company the normal way, you basically create a community first and then you take that community and use the community to design the future structure. So, at the stage that we are at with Dreamblocks, which is a relatively new company, is that we have created the initial community and we are due to have a meeting to start thinking about how the company is going to be structured. You also asked about NFTs, which I am very familiar with, because I am also an artist. Although my art is not normally digital, like most artists I have spotted the potential of NFTs, which is to add value to digital art, so I have been experimenting with digital art and NFTs. But I am also interested in some of the wide aspects of NFTs. How do DAOs come into Blockchain & what do they do? PD:| I think DAOs are one of the potential use cases of Blockchain. So Blockchain is really the underlying infrastructure that drives like Bitcoin, but it also has this second ability which is to be able to decentralize organizations. And what you can do with Blockchain is you can create smart contracts which themselves are the bases of DAOs so there is an integral linkage between Blockchain and DAOs. cryptocurrency How does AI & Data science fit into NFTs & Blockchain? PD:| Well I think when we talk about AI and in their own right and what we are looking at is the intersection between the two. I is that we are utilizing Data in increasing and maybe invasive ways, that could be easily manipulated by people or end up in the hands of the wrong people. Blockchain we are getting into the most advanced use cases because what we are talking about is two very complex technologies think it is very important because what is happening with Artificial Intelligence So, I think where Blockchain starts to come in, this is not directly or necessarily linked to , but it will be in some cases, but more to the point it’s a way of dealing with people’s Data in a more equitable manner, in a way in which they could be rewarded for their own Data. Machine Learning For example, you can own your own Health Data and, even though owning your own health Data is not necessarily a use case for AI in its own right, it’s only your own Data, what can then be done is that you can provide your Data to someone who can then put on top of it to create insights and be able to discover new drugs, for example, and things like that. Machine Learning Where will DAOs be in 5 years from now? PD:| So I think that there are very complex ideas and very complex visions around what Web 3.0 should look like which are quite easy to talk about but quite hard to implement. Well, if you look at talks from Davos yesterday or if you are looking at something like Polka Dot which is a Blockchain they are really talking about how we avoid the mistakes that have been made in Web 2.0. So, what is happening with Web 2.0 is we have centralized like Facebook, Instagram and all the Web 2.0 companies. We have got the advantage of seeing what went wrong with Web 2.0 and now what we need to do is we need to use these new structures to do it in a much more equitable way but it is easier said than done so I think to answer your questions where will it be in 5 years time? Data and we have given the power of the Data to the people who own the company Well in five years’ time we would have ironed out some of these problems that will allow us to be able to somehow balance questions like freedom of speech with flexibility and speed and all the things we take for granted with Web 2.0 and we would have figured out how some of that technology works in practice because I think at the moment we are in the very very early stages just taking little baby steps with these technologies. What are the benefits of DAOs? PD:| For me the benefits are that they allow a more equitable structure that allows more people and more stakeholders to be rewarded in a more even way. So again when you compare it to Web 2.0, the problem with Web 2.0 at the moment is that a founder is allowed to make a lot of money by his venture capital and the venture capital will make a lot of money as well. There is a symbiotic, biotic relationship between the VC and the founder that says let’s go out and make as much money as we can, but what happens is that when they write the contracts for the rest of the team and so on and so on it gets less and less extractive. When you get right down to, for example, the drivers on Uber, by the time it gets down to them they are just employees and they have no real ownership of the startup. I think what DAOs do is that they bake in the rules of the game right from the very beginning and they don’t allow those rules to change, except in very exceptional circumstances where there is some kind of agreement, so it doesn’t allow the founders to slide out of their commitments as the company develops and actually gets better as a company develops because of all the voting mechanism. So, in the case of a Web 3.0 Uber, the drivers would have more power. And the other thing is, that it bakes in the incentives mechanisms, so right from the very beginning, it says if you are a driver and follow this kind of behavior you get a little more of the tokenized tokenization through tokenization and therefore you are rewarded for good behavior as the company grows. How do you set up a Decentralised Autonomous Organisation? PD:| Right! Well, the questions are getting increasingly hard, aren’t they? Well, it is hard for most people to understand but especially hard for me to understand. So there are mechanisms for creating DAOs, I mean they are really effectively a legal structure and then lawyers are beginning to understand more and more how they can be set up. A DAO really has to be designed by the community itself. I think people would like to pay lip service to DAOs and would like to gather a small team, just like a normal management team, and create a DAO; but that just would be another way of creating a team in the old sense. So I think to create a DAO in a true sense I think you probably have to think of the mission you are trying to tackle and it could be anything from, let’s just take an easier one: Climate Change. You should first build your community around Climate Change and should tell them in the future you will be creating a DAO, but you need to get some consensus before you start to design the DAO and then you need to take that community or a subset of that community, and that needs to be the community that designs what the DAO looks like. Because a DAO really is a set of rules basically: if this happens then that will happen and those rules need to be baked in from the beginning. How much does it cost to set up a DAO? PD:| I haven’t a clue! Ha, ha. Is NFT worth investing in and where do you think it will reach in the near future? PD:| I think first of all there is an artist’s view of this and then I think secondly there is a business view of this which is different. I think from the artist’s perspective, at a really really simple level what NFTs have allowed to happen, for digital artists is to monetize their work in a way that has not been possible before. But it doesn’t that, like all art, it’s only a very tiny parentage that has artistic merit. solve the problem So it does not suddenly give you artistic merit that you didn’t have before, but what it does is, let’s say out of 100 digital artists, like out of 100 normal artists, there will be 5,1 or 2 that create great art and now, whereas before it would have been shared randomly round the Internet and they would have not got many rewards from it, now they can actually create rarity for that art and, because of NFTs will now start to have true value and high value. But what a lot of people are believing is that the other 90 people who got no talent can also suddenly create value, which actually at the moment is just speculation value, so it is really people who don’t understand that. But it is slightly better than that, because what it also does is that it allows you, whether you got talent or not, to use the NFT effectively as a token. So as long as you add enough utility to your artwork it could be the artwork is worth 0 really but the utility is actually worth something on top of it and you just use the NFT as a token. So, coming back to the Climate Change example, you could create a set of 20 Climate Change images that may not particularly have huge artistic merit, but people might buy those because they believe in the cause. Because if you say every NFT is worth £50, if you buy one of those you are giving £50 to Climate Change NFT, which goes directly into the utility. And that utility can be something serious like Climate Change, but it can also be something quite frivolous, like entertainment. It could be your a rapper and you want someone to buy your NFT, so again you might not be buying it because of its merit, you might be buying it partly because of its merit, maybe it is a music NFT so maybe you are buying it because of its music, but actually, you are also buying it because whatever else the raper has added to it. So, he might add that you get the right to meet them at every concert, or you have the right to come to every concert without paying. And you can trade that, and that’s the other good thing about NFTs, that they have a secondary value that once you bought them, then other people can buy whatever utility is in that NFT and therefore they are tradable. When you think of a standard piece of artwork once the artist sells it, that’s it. They can’t keep tracking that piece of art when it is worth millions they can’t and say I was the original artist, give me 10% of that million but with NFTs it does. come back What’s the connection between NFTs and cryptocurrency? PD:| My gut feeling is to say that the connection should be somewhat avoided, I mean they are enabled by the same technology and there is a risk, I suppose, that they might get too associated, for example, they could become a speculative item, they could become very close to a currency. In other words, if somebody knows that a certain type of then it becomes just like a currency, but that should not really be the aim of an NFT. I think the aim of an NFT should be almost the opposite to that. It should really be about what the artistic merit or utility of it should really be the thing that prevails, rather than just seeing it as another currency. NFT is always going to go up in value What makes an NFT valuable? PD:| So I think it is artistic merit in its own right, it can have a huge value. We all know great art is worth a lot of money. Or it is the extra utility in it. Or if you are really lucky it can be both artistic value and extra utility. How do NFTs work in properties and decentralized home rentals? PD:| So first I will dissect this question a little bit, because I don’t think it is just NFTs. We can say this is more about Blockchain and there are two use cases that are separate. So actually one thing that is becoming very large is real estate on the , so I think this is where more NFT comes in. metaverse So the NFT could be a token to buy in the Metaverse which is already happening. I met with a company that has already sold £8m of Metaverse real estate. But then of course there is the use of Blockchain and Smart Contracts to take the middleman out of the property buying process and that’s probably the most exciting one really, because property is a great example of having to pay a bunch of people a lot of money, for really doing very little, just because it is a complex process, and up to now requires middlemen to play a part in it because of the questions of trust and things like that. You and I can buy a house but we are not really allowed to do that, because there is no automation, or there is no validation and trust built into that process. real estate So we go and get lawyers and we get state agents and others to whom we have to pay a very high percentage before we can buy a house. The theory would be if you put real estate, either rentals or sales, on to Decentralised Models with Smart Contracts: if you did A and I did B then the contract would be sealed by the Smart Contract. It will basically track what we do and it would be able to validate things like ownership of the land because that would all be put on the Blockchain. It would not need a surveyor to come and say “you own this land”.  It will just be there on the Blockchain, a unique piece of land you own and then that enables the process of buying and selling. Also Published Here

Mindstream AI CEO Paul Dowling Talks About DAOs, NFTs and the Future of Web3.0

by Anne Szustek Talbot, director of content, BX3 Capital From the vantage point of cryptocurrency players and analysts, Congressional action on the part of cryptocurrency will likely stay on its same general course, albeit with the possibility for a more decidedly Democrat flavor: how regulation stands to affect the retail-level consumer as much as the big-time asset management firm. “The members of Congress who are spearheading the crypto regulation initiatives all got re-elected, so the sponsors behind the bill will likely remain the same,” says Kyle Asman, partner and co-founder at , and one of the crypto industry members who testified in front of the Congressional roundtable. “But in order for the bill to make it out of the House Financial Services Committee and onto the floor, the now-left-leaning committee will likely make consumer protections a top priority.” BX3 Capital Warren Davidson, the Ohio Republican who spearheaded the , was re-elected, as were his GOP roundtable colleagues and , both of whom have drafted cryptocurrency bills since the September talk. Emmer is also one of the four co-chairs of the 18-member , most of whom will be returning in January. Co-chair Jared Polis (D.-Colo.) is staying in the Rocky Mountain State as the . (What will this mean for crypto in Colorado?) Longtime Virginia , to be replaced by fellow Republican Ben Cline. Pennsylvania Republican Keith Rothfus succumbed to the House “blue wave,” as appears will Tom McArthur as the votes get counted in his tight race against , which stretches from outside of Philadelphia to the Jersey Shore. September Congressional roundtable Ted Budd (R-N.C.) Tom Emmer (R-Minn.) Congressional Blockchain Caucus nation’s first openly gay governor Republican Bob Goodlatte is retiring Democratic challenger Andy Kim in New Jersey’s 3rd Congressional district Minor shifts among Congress’s top cryptocurrency brass or not, beyond any possible consumer-watchdog tacks, don’t expect major sway in crypto policy. From the early days of Congress’ cryptocurrency task force, crypto has been a bipartisan cause that has remained far from being a hot-button topic for rhetoric. “Neither political party made the topic a campaign issue, as the space is still new and generally does not affect the average taxpayer,” says in the securities practice at the Los Angeles office of law firm Drinker, Biddle & Reath. JR Lanis, partner and cryptocurrencies expert include H.R. 3708, a bill to exempt small transactions (less than $600 at inception, pegged to inflation going forward) from IRS reporting, as well as H.R. 6973, meant to protect taxpayers from regulation by enforcement on convertible cryptocurrency received following a blockchain forks. Cryptocurrency bills pending in Congress H.R. 6974 also seeks to provide a regulatory safe harbor for crypto, but on the licensing and registration front for blockchain providers and developers that do not “control” users of the software of the network. For the purposes of the bill, “control” is defined as “the legal right, authority, or ability to obtain upon demand data sufficient to initiate transactions spending an amount of digital currency.” David Schweikert (R-Ariz.), another of the Congressional Blockchain Caucus co-chairs, is the lead sponsor of H.R. 3708; Emmer is heading H.R. 6973 and 6974. Budd is the leading sponsor of H.R. 6721, a bill that would provide that the gets monitoring technology for cryptocurrency transactions and other novel financial technologies. Each of these bills are still in committee; whether or not they reach the House floor for a vote is largely dependent on factors beyond political winds. Financial Crimes Enforcement Network The consensus in the cryptocurrency industry is that regulation needs to happen — no matter which side of the aisle gets it done — to prevent more regulation by enforcement, as two of the pending bills seek to avoid. Participants at the September roundtable were nearly unanimous in mentioning ambiguity as a reason for taking cryptocurrency-based businesses overseas, lest they make the wrong step without knowing they took a wrong step in the first place. The pending bills will likely survive the Congressional term next January, though with the chance for some slight tweaks to language. In the meantime, members of Congress have asked the SEC for some clear guidance on initial coin offerings (ICOs). As far as Congress is concerned, the cryptocurrency sector can rest somewhat easy. As has been the case for some time now, it’s the financial enforcement bodies where crypto watchers need to focus their attention as we head into 2019. Even on this front, there may be softer times ahead. Says Landis, “U.S. regulators, especially the SEC and CFTC, have demonstrated an increasing recognition that new rules and guidance must be issued — as opposed to relying on old paradigms — to address this growing space.”

How We Think Congress Will Treat Cryptocurrencies After the 2018 Midterms

Can AI improve midterms forecasts? Context awareness AI is larger than machine learning. The physical world around us and our society compute themselves, constantly. What is the use of glorified statistical models that tell us something about how the world was yesterday -what happened in the last election- if our machines do not perceive what is important to humans today, the sudden & subtle changes in human priorities? People vote with ballots and with their wallets People really care when they put money on the table -it is unlikely that a voter who was materially committed to a cause by means of a political donation does not show up to vote. That (a driver) that machines can count on. But is also possible that after outspending on amplification, the contest is lost because resources were misallocated. Luckily, once provided with purpose, machines excel at planning. Therefore, political campaigns need to understand those trust signals and . measurement of trust is a signal optimization objectives Online contribution platforms We study the user behavior in online contribution platforms, to discover what campaigns are more/less successful -then we study those campaigns individually to learn they are successful. why Since this is a quantitative macro approach based on open data, we do not access individual private data at any point — we are concerned only about group behavior and observed probabilities (what actually happened). The data is generated for specific periods of time (e.g any month going back 2 years), or on what is a nearly real time in the realm of campaign finance (2 days ago to 4 trailing weeks). Weighing for recency allow us to account for the effect of momentum. MyNGP One good example to look at is the NGP platform. The popularity of campaign contributions The figure shows thousand of campaigns and movements that collect donations. Some may capture a large share of attention at a moment in time, but do not sustain grow. Some are evergreen: they generate growth outbursts many months after they first started. Most are too little to likely make any long-lasting difference. And there are the whales: the candidates or causes that account for dozens of the small ones. But how? You can find all sort of surprises. For instance, democrats use republicans overtures to advance their own narratives, like in the case of one of the top donation pages in Actblue (Actblue is another fundraising platform frequented by democrats) that is not about a democrat but about a cause, and actually uses the position of a former republican presidential candidate as a pitch for “tax fairness”. Other services, such as Anedot, are popular among GOP candidates and are worth studying to develop a digital campaign strategy that covers multiple angles. By having this information campaigns can compare their performance to the most successful within their own party, but also between competing political parties — then, replicate the tactics that work. Augmenting human intelligence Often times, traditional methods such as polling and history of fundraising are not definitive — a few seats are too close to call. It can even happen that “expert” forecasts and comparable-based statistical methods are completely off. In those cases the of AIs that sense and compute the -of people who actually pay attention and commit resources- is akin to a professional punter. embedded intelligence belief consensus Because even Nate Silver has to toss-up a coin, sometimes. FiveThirtyEight Fivethirtyeight, ABC The rise of decentralization in politics One of the main takeaways is to see the very low traffic in the majority of the small district campaigns — they have no forceful online media strategy. In general, pages where the candidates branding is strong and the value proposition/message/imaging is loud & clear, perform well. Pages that simply ask for money upfront do less well. A bad way to ask for money, is to do nothing more than asking for money But there is something more fundamental going on: . One can see it with the prevalence of peer-to-peer (P2P) texting; the bitcoin community rally behind libertarians, greens, and Sanders during the past election; the emergence of mobile Apps as social media alternatives to convey a discourse and to build grassroots movements. the decentralization of political financing A is particularly good at doing cohort analysis that shows the “churn” patterns of the users of popular political apps. With these insights you can understand when you will “lose” the attention of your voter, and when the re-engagement activities should be programmed. For instance, political risk AI 1.You can capture the attention of voters that are interested in your competitor. 2.You can the attention of someone who visited your fundraising platform, even if they did not committed funds. re-capture Having this information on hand, operational implementation is straightforward: campaigns can engage person to person in the places where THEY ARE, or use programmatic marketing services to display ads in those places specifically (e.g. where & when they consume news). Without this knowledge, you will be shooting in the dark, and burning promotional budget. Case in point: A candidate that did not win the primary in his congressional district, but received 80,000 online donations. Desktop users of his website on a particular month were mostly male (51.84%), and over 65 years of age (22.61%). In the other hand, mobile users were predominantly female (59.2%) and between 55–64 years old (22.84%). Younger generations (millennial, etc) were a small fraction in both cases. This difference matters because that month the website received 72.89% of visits from mobile devices, which indicates that there was a “blind-spot” in terms of reach — or perhaps an opportunity to engage that population (around 19.1 thousand women) as standard bearers. The candidate had a strong personal brand, 60.44% of the visits from search engines such like as Google came from the candidate’s name query (around 4,630 visits), and it would cost him $ 0.34 per click to get those visits using paid ads. Another sign of his strong candidate brand is the fact that 81.14% of the visits to the website were direct (not coming from email, or search), and more than 70% of the visitors checked more than 1 page before leaving the site. In his case, fundraising was only half way through: he needed the intelligence data to understand where his political efforts were at risk. The digital realm is a competition for attention — if people are interested in local politics, but are not spending time at your site or at your competitors, they are probably hanging out where the undecided are. And today, that happens to be mobile, the handful of apps that people use repeatedly, and their cross-app usage in the intersection of life and politics. Based on a political risk forecasting algorithm developed by https://www.economymonitor.com EM’s Percy Venegas is also the author of , published by . Trust Asymmetry Springer

AI & the Midterm Elections

Machine is kind of magic right? But is it the kind of magic that can make us rich? And I don’t mean lucrative consulting gig rich, I mean digging valuable metals out of the ground rich. The kind of rich you can horde in treasure chests. learning Also I’d been meaning to try out some transfer learning and looking around for a good topic to try it on. Transfer learning is where you take a pre-trained convolution (or other) and use it for your task. It’s great for convolution networks particularly because they take a long time to train. South Australia where I live has taken a lot of effort to share data sets, including some great geological resources, so I thought why not stick the two together and use convolution networks to find gold. The easiest way to make use of convolution networks is to turn them into feature extractors, which is what I’ll show you in this post. network The code that goes along with this blog post can be found on GitHub: _DeepGold using convolution network features to learn mineral data_github.com scottvallance/DeepGold Minerals So to have a computer find you treasure, you’ve got to give it something to work with. Like a treasure map. So my first assumption is that a map of geological features might be enough to spot likely places if you knew how to look at it. Of course it’s not like geologists don’t do this already. They probably even know the names of the minerals and whatnot’s they are looking for unlike me. So I found some papers written on finding minerals (gold in particular) using images. Instead of my first idea of maps, which would have been a fun challenge for a convolution network, there is a multi-spectral satellite image source called ASTER which can provide clues to mineral resources on and in the ground. _ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) is a high resolution imaging instrument that is…_asterweb.jpl.nasa.gov ASTER: Advanced Spaceborne Thermal Emission and Reflection Radiometer Multi-spectral imaging is taking a picture with a camera that can see more than just red, green and blue. Some of these bands in ASTER are good at spotting particular minerals, or mineral families. This following paper goes into how these bands have been used previously and how they could be used. _Remote sensing plays an important role in mineral exploration. One of its proven applications is extracting and…_www.researchgate.net Detecting areas of high-potential gold mineralization using ASTER data While the paper comes up with their own scheme, they mention the following mineral indexes as being particularly good for finding gold: “OHI is the index for OH-bearing minerals, KLI is the kaolinite index, ALI is the alunite index, and CLI is the calcite index”. ASTER has a bunch places where we can download it free of charge, but the downside of wonderful real data set like ASTER is that it’s full of all sorts of real world complications. Like clouds getting in the way, and like the fact that since 2008 some of the sensors on the satellite are basically cooked. Fortunately Geoscience Australia have done a lot of the hard work of stitching and manipulating the data and produced a bunch of big ‘.tif’ images to work with. _This Australian ASTER Initiative, led by CSIRO's Western Australian Centre of Excellence for 3D Mineral Mapping…_www.ga.gov.au National ASTER Maps - Geoscience Australia You can find the data on their ftp site with the links in the Github repo for this post. I grabbed a bunch of images for the location I was interested in, using the AlOH, MgOH and Kaolin band images. That’s only half the story though, we have a lot of images but no clue where the gold is. The easiest way to use machine learning to help us find gold is to give it a bunch of images that have gold in them and a bunch that don’t and get it to figure out the difference. To get that training data we need to know where there’s gold. Another government department the rescue, this time a South Australian state department which has a great little tool they call SARIG. _Edit description_sarig.pir.sa.gov.au South Australian Resource Information Geoserver - Map It’s a map with a lot of different mineral related overlays, one of them particularly interesting to us — ‘All Mines and Mineral Deposits’ with a handy option to download the shape file geo database we have everything we need data wise. Data Processing Now we have lots of data, but it’s not simply a matter of feeding it into computer. You have to slice and dice it into the formats needed. To work geospatial data their are a bunch of great tools collectively known as GDAL: Geospatial Data Abstraction. _Traditionally GDAL used to design the raster part of the library, and OGR the vector part for Simple Features. Starting…_www.gdal.org GDAL: GDAL - Geospatial Data Abstraction Library These tools will happily export info out of our ‘All Mines and Mineral Deposits’ shape files and slice up the massive image files that are also geospatially aware. As each mineral deposit entry in shape file is a point location, we can cut up the large input images into segments around each point of a size appropriate to feed into a convolution network. These slices can have multiple minerals and we don’t allow overlapping slices so that the eventual machine learning doesn’t just get to remember what different segments look like and learn from rote. Then it’s a matter of merging the images from the different bands into a false colour image. A false colour image is when we use something other than real red, green and blue values in image — in this case we’re using the AlOH content, the MgOH content and the Kaolin index. One measure for each colour band. Here’s what a false colour image ready to feed into our convolution network looks like: A false colour image using GEOScience Australia reprocessed ASTER data Convolution Networks Convolution networks have shown amazing ability to learn answers to visual problems. Recently there’s a been a bunch of open source tools released to develop convolution networks including: TensorFlow, Theano, Caffe, Torch. All great tools but we really need something with easy access to pretrained data. For that problem, and for abstracting convolution networks a step away from their implementation, Keras has been incredibly useful. _Keras: Deep Learning library for Theano and TensorFlow_keras.io Keras Documentation The author also provides this with models and pretrained data to work with. The has produced good results, so that’s an obvious starting point for the network. We’ll be using this network to transform our training images into a vector features. Essentially we take the output of the ResNet50 at a point before it turns into a bunch of image class predictions and use that as the input for another classification problem. While training a convolution network is slow using it to generate features is quick and gives us a data source that has been proven to transfer to other problems well. It’s definitely an open question how well it transfers to something as esoteric as a processed false colour satellite image but trying it out is the easiest way to learn. Github repo ResNet50 network Classification with ResNet50 Features After we turn our false colour images into features we can run a bunch of different machine learning algorithms. For this we make use of Spark, a cluster focused graph computation framework with a number of built in machine learning algorithms. _Apache Spark is a fast and general engine for big data processing, with built-in modules for streaming, SQL, machine…_spark.apache.org Apache Spark™ - Lightning-Fast Cluster Computing While the size of the problem doesn’t necessitate using a cluster to solve it, it does provide a nice interface to the algorithms. We can make use of Logistic classification, Support Vector Machine classification, Naive Bayes classification, Random Forests and Multilayer Perceptron classification with just a few lines of code. To make our data ready for Spark we turn it into LibSVM format and then further simplify it by filtering the examples into those containing gold and those that don’t. This leaves us with a lot more not gold examples, so we over-sample the gold ones to give us even numbers because it helps to train an algorithm with balanced data. We load up pyspark and use a few lines of code to run some different classifers. Here are the results of Spark’s classification algorithms: Logistic Regression:Test Accuracy = 0.767756482525 SVM:Test Accuracy = 0.808342728298 Naive Bayes:Test Accuracy = 0.616685456595 Random Forest (10 trees):Test Accuracy = 0.727171 Multilayer Perceptron (layers [2048, 512, 2]):Test Accuracy = 0.775255 This is just a measure of how often the algorithm was right about an image containing gold or not over the total number of testing examples. An accuracy of 80% doesn’t sound bad, and does match with other results shown in the literature but it’s difficult to interpret this data from a simple accuracy measure. We haven’t over sampled the gold examples in the testing data, so they contain a lot more not gold than gold. If an algorithm gets pessimistic about predicting gold it can still do quite well while not telling us very much about where to dig. What we really need is a measure of how likely we are to strike gold if we do go out with a shovel. This is called the precision of the prediction. The flip side of an algorithm being really precise is that it may choose to only predict gold when it’s very sure and miss a lot of examples. The measure of the amount of examples we correctly predict as gold out of the total gold samples is called the recall. We’re probably fine with a low recall, but we’d like precision to be as high as possible. Here’s the precision and recall for the data above: Logistic Regression:Gold Recall = 0.324840764331Gold Precision = 0.337748344371 SVM:Gold Recall = 0.235668789809Gold Precision = 0.425287356322 Naive Bayes:Gold Recall = 0.68152866242Gold Precision = 0.269521410579 Random Forest (10 trees):Gold Recall: 0.575163398693Gold Precision: 0.328358208955 Multilayer Perceptron (layers [2048, 512, 2]):Gold Recall: 0.317880794702Gold Precision: 0.335664335664 So for the purposes of getting out there with a shovel the Support Vector Machine classification really wins. If it says there’s gold then there’s a 42% chance it’s telling the truth, which is out of 17% of samples in the testing data actually having gold. This is just for segments of South Australia that have known mineral deposits though, which is a factor to consider as it might correlate to certain features that aren’t seen in other areas. ‘X’ Marks the Spot So given the data we have is where do we go digging? Obviously not where gold is already labelled because it’s already dug up. Which really just leaves where it isn’t labelled, which has the down side of already being mined for something — but not gold. We’ll take it though, a lot of the data in the mineral deposits is old and abandoned. The Support Vector Machine performed best, but we want to move beyond classification into probabilities. If we clear the threshold that turns the SVMModel in Spark into a classifier we can get raw scores straight from the estimates. Taking the max of the ones where there is no gold gives us this image: Weird boomerang shape marks the spot Let’s check it on the map: Our great mining hope Someone grab the shovels, we’re off! Turns out this is right next door to the White Dam Gold Mine. Which is I guess a good sign, but the mine is actually in our data already it’s just a bit to the left and down. There’s a good chance that the spot we found looks a lot like the mine next door. A quick look through the rest of the likely spots shows a lot of regions next to existing mines. Probably a good thing, but sort of hard to make use of. Running through the maps without mineral data section by section might yield something but it’s a bit speculative and might best be left to the reader. So we didn’t really find gold, but I’m still pleasantly surprised by the accuracy of convolution network features on such a different problem to that which they were trained. Conclusion and Other Projects After searching through the maps with the algorithms we’ve developed a great next step would certainly be to train a convolution network on raw ASTER data. Convolution networks are getting to human levels of success (and over) on certain vision tasks and I think it they could have a big advantage with multispectral data. Humans can only interpret visual data through at most (theoretical not withstanding) three channels but convolution networks are not similarly constrained. The current band ratio used for finding minerals would certainly map well to a network architecture. tetrachromats If you find this article interesting you can . find me on linkedin

Deep Gold: Using Convolution Networks to Find Minerals

The Quiet Crisis That Is Stealing Our Health (and How to Get It Back) - Insights From Dr. Eric Berg

HyperTransformer: Conclusion and References

So, let’s start! Tell us a bit about yourself. For example, name, profession, and personal interests. I’m João, and I have been a Software Engineer for more than 10 years now. I am technically a Software and Telecom engineer, but most part I have only been doing software engineering professionally. There are many things I love to do, and that, of course, includes making Tech videos and writing blog posts and articles, and maintaining my website. Outside work and outside Tech, I am someone very curious about biology, which is why I like to try to maintain a garden where I grow things like blueberry bushes, honeysuckles, and gooseberry bushes, and I also love photography. I mostly enjoy photographing nature, cities, and things with lights. Further, I also like to enjoy my time doing stuff related to art, for example, video, photo editing, and style editing. Outside of all of this, I also enjoy music, series, films, traveling, cooking, and, from time to time, the occasional electronics project. Interesting! What was your latest Hackernoon Top story about? My latest Hackernoon Top story was the one I did about Kuma, which is about a way to create Meshes using Kuma-Meshes, where we can have a much more granular control over the data flow within our own network, which, amongst other things, can protect an organization against internal bad actors. Do you usually write on similar topics? If not, what do you usually write about? A the moment I exclusively write about Tech, although I can form time to time relate those posts with other kind of topics like opinionated arguments or even politics when they have the potential to influence tech. It can also be that I can write about how some particular tech stacks may influence our approach to tech or even society and environment. But the main focus is always tech. Great! What is your usual writing routine like (if you have one?) I do not have a specific writing routine, although I am writing a lot more these days. Being a writer in tech can be a challenge. It’s not often our main role, but an addition to another one. What is the biggest challenge you have when it comes to writing? My biggest challenge in writing is creating a functional example as fast as possible. Creating a story, implementing needed elements, making them run, creating tests for it, creating databases or any other persistence mechanisms for it, and trying to guarantee as much as possible that people can use the examples without walking against errors or faults is usually a very energy-intensive work. Therefore, it can take longer than desired. What is the next thing you hope to achieve in your career? I want to be as successful as possible with my own effort and work, and I want to share my experiences and knowledge with other colleagues in the world of software engineering. I don’t know if it eventually can become a full-time job, but if it does, it will probably be the next thing that I would love to achieve. Wow, that’s admirable. Now, something more casual: What is your guilty pleasure of choice? Listening to Madonna’s Ray of Light album. Do you have a non-tech-related hobby? If yes, what is it? Yes! I love to take photos of wildlife. I especially love making macro photos of leaves, insects, and flowers, as well as minerals, rocks, glass, metals, and anything that looks outstanding in a macro photo. What can the Hacker Noon community expect to read from you next? I usually write more about topics related to the Kotlin programming language; however, having said that I am finding it more and more interesting to write about security and operations from the DevSecOps area as I did in my article about Kuma. I think we can expect to read more about the SecOps area in my next articles, but I will still write about Dev. What’s your opinion on HackerNoon as a platform for writers? I think HackerNoon is a unique platform for writers to share knowledge about Technology because it is very recognizable, and a lot is automatically done for us like, for example, in this last article when it comes to translating my Kuma article in multiple languages. That was a very nice surprise to see. All in all, I think HackerNoon really appeals to Software programmers, Engineers, Developers, and even Businesses and so I’m very happy to keep writing for HackerNoon. Thanks for taking time to join our “Meet the writer” series. It was a pleasure. Do you have any closing words? Yes, thank you for giving me this great interview. I really appreciate the boost given to my article. I hope to be able to keep writing for HackerNoon for as long as possible. It is a pleasure to do so. Thanks again for the incredible support and for being there. So, let’s start! Tell us a bit about yourself. For example, name, profession, and personal interests. I’m João, and I have been a Software Engineer for more than 10 years now. I am technically a Software and Telecom engineer, but most part I have only been doing software engineering professionally. There are many things I love to do, and that, of course, includes making Tech videos and writing blog posts and articles, and maintaining my website. Outside work and outside Tech, I am someone very curious about biology, which is why I like to try to maintain a garden where I grow things like blueberry bushes, honeysuckles, and gooseberry bushes, and I also love photography. I mostly enjoy photographing nature, cities, and things with lights. Further, I also like to enjoy my time doing stuff related to art, for example, video, photo editing, and style editing. Outside of all of this, I also enjoy music, series, films, traveling, cooking, and, from time to time, the occasional electronics project. Interesting! What was your latest Hackernoon Top story about? My latest Hackernoon Top story was the one I did about Kuma, which is about a way to create Meshes using Kuma-Meshes, where we can have a much more granular control over the data flow within our own network, which, amongst other things, can protect an organization against internal bad actors. Do you usually write on similar topics? If not, what do you usually write about? A the moment I exclusively write about Tech, although I can form time to time relate those posts with other kind of topics like opinionated arguments or even politics when they have the potential to influence tech. It can also be that I can write about how some particular tech stacks may influence our approach to tech or even society and environment. But the main focus is always tech. Great! What is your usual writing routine like (if you have one?) I do not have a specific writing routine, although I am writing a lot more these days. Being a writer in tech can be a challenge. It’s not often our main role, but an addition to another one. What is the biggest challenge you have when it comes to writing? My biggest challenge in writing is creating a functional example as fast as possible. Creating a story, implementing needed elements, making them run, creating tests for it, creating databases or any other persistence mechanisms for it, and trying to guarantee as much as possible that people can use the examples without walking against errors or faults is usually a very energy-intensive work. Therefore, it can take longer than desired. What is the next thing you hope to achieve in your career? I want to be as successful as possible with my own effort and work, and I want to share my experiences and knowledge with other colleagues in the world of software engineering. I don’t know if it eventually can become a full-time job, but if it does, it will probably be the next thing that I would love to achieve. Wow, that’s admirable. Now, something more casual: What is your guilty pleasure of choice? Listening to Madonna’s Ray of Light album. Do you have a non-tech-related hobby? If yes, what is it? Yes! I love to take photos of wildlife. I especially love making macro photos of leaves, insects, and flowers, as well as minerals, rocks, glass, metals, and anything that looks outstanding in a macro photo. What can the Hacker Noon community expect to read from you next? I usually write more about topics related to the Kotlin programming language; however, having said that I am finding it more and more interesting to write about security and operations from the DevSecOps area as I did in my article about Kuma. I think we can expect to read more about the SecOps area in my next articles, but I will still write about Dev. What’s your opinion on HackerNoon as a platform for writers? I think HackerNoon is a unique platform for writers to share knowledge about Technology because it is very recognizable, and a lot is automatically done for us like, for example, in this last article when it comes to translating my Kuma article in multiple languages. That was a very nice surprise to see. All in all, I think HackerNoon really appeals to Software programmers, Engineers, Developers, and even Businesses and so I’m very happy to keep writing for HackerNoon. Thanks for taking time to join our “ Meet the writer ” series. It was a pleasure. Do you have any closing words? Meet the writer Yes, thank you for giving me this great interview. I really appreciate the boost given to my article. I hope to be able to keep writing for HackerNoon for as long as possible. It is a pleasure to do so. Thanks again for the incredible support and for being there.

Meet the Writer: HackerNoon's Contributor João Esperancinha, Software Engineer

Authors: (1) Philipp Straubinger, University of Passau, Passau, Germany and this author contributed equally to this research; (2) Tommaso Fulcini, Politecnico di Torino,  Torino, Italy and this author contributed equally to this research; (3) Gordon Fraser, University of Passau, Passau, Germany; (4) Marco Torchiano, Politecnico di Torino, Torino, Italy. Table of Links Abstract and Introduction Background and Related Work Implementation Experiment Results Conclusions, Acknowledgement, and References 6 CONCLUSIONS This paper presents the integration and empirical evaluation of IntelliGame, a gamification plugin originally designed for Java development, in the realm of JavaScript unit testing. The study aimed to validate IntelliGame’s effectiveness in JavaScript, adapting it to the popular Jest testing framework. Despite challenges such as transitioning from IntelliJ Community Edition to Ultimate for JavaScript support and addressing differences in coverage information, we successfully ported 19 out of 26 achievements. The controlled experiment with 152 participants revealed mixed perceptions of IntelliGame’s impact, with achievements influencing testing behaviour and participants’ motivation. We will continue with an in-depth analysis of the experimental measures and the code written by the participants to gain more insights and to broaden our knowledge of IntelliGame. ACKNOWLEDGEMENT This study was carried out within the “EndGame - Improving Endto-End Testing of Web and Mobile Apps through Gamification” project (2022PCCMLF) – funded by European Union – Next Generation EU within the PRIN 2022 program (D.D.104 - 02/02/2022 Ministero dell’Università e della Ricerca). This manuscript reflects only the authors’ views and opinions and the Ministry cannot be considered responsible for them. This work is also supported by the DFG under grant FR 2955/2-1, “QuestWare: Gamifying the Quest for Software Tests”. REFERENCES [1] Riccardo Coppola, Tommaso Fulcini, Luca Ardito, Marco Torchiano, and Emil Alègroth. 2023. On Effectiveness and Efficiency of Gamified Exploratory GUI Testing. IEEE Transactions on Software Engineering (2023), 1–16. https://doi.org/ 10.1109/TSE.2023.3348036 [2] Sebastian Deterding, Dan Dixon, Rilla Khaled, and Lennart Nacke. 2011. From Game Design Elements to Gamefulness: Defining "Gamification". In Proceedings of the 15th International Academic MindTrek Conference: Envisioning Future Media Environments (Tampere, Finland) (MindTrek ’11). Association for Computing Machinery, New York, NY, USA, 9–15. https://doi.org/10.1145/2181037.2181040 [3] Tommaso Fulcini, Riccardo Coppola, Luca Ardito, and Marco Torchiano. 2023. A Review on Tools, Mechanics, Benefits, and Challenges of Gamified Software Testing. ACM Comput. Surv. 55, 14s, Article 310 (jul 2023), 37 pages. https: //doi.org/10.1145/3582273 [4] Andreas Jedlitschka and Dietmar Pfahl. 2005. Reporting guidelines for controlled experiments in software engineering. In 2005 International Symposium on Empirical Software Engineering, 2005. IEEE, 10–pp. [5] José Miguel Rojas and Gordon Fraser. 2016. Code Defenders: A Mutation Testing Game. In 2016 IEEE Ninth International Conference on Software Testing, Verification and Validation Workshops (ICSTW). 162–167. https://doi.org/10.1109/ICSTW. 2016.43 [6] Statista. 2023. Most used programming languages among developers worldwide as of 2023. https://www.statista.com/statistics/793628/worldwide-developersurvey-most-used-languages/ Accessed: 2023-12-05. [7] Philipp Straubinger and Gordon Fraser. 2023. A Survey on What Developers Think About Testing. In 2023 IEEE 34th International Symposium on Software Reliability Engineering (ISSRE). 80–90. https://doi.org/10.1109/ISSRE59848.2023.00075 [8] Philipp Straubinger and Gordon Fraser. 2024. Improving Testing Behavior by Gamifying IntelliJ. In Proceedings of the 46th IEEE/ACM International Conference on Software Engineering, ICSE 2024, Lisbon, Portugal, April 14-20, 2024. ACM, 49:1–49:13. https://doi.org/10.1145/3597503.3623339 [9] Kwangwon Sun and Sukyoung Ryu. 2017. Analysis of JavaScript programs: Challenges and research trends. ACM Computing Surveys (CSUR) 50, 4 (2017), 1–34. [10] Bo Wang, Aashish Kolluri, Ivica Nikolic, Teodora Baluta, and Prateek Saxena. 2023. User-Customizable Transpilation of Scripting Languages. Proc. ACM Program. Lang. 7, OOPSLA1 (2023), 201–229. https://doi.org/10.1145/3586034 This paper is under CC BY-SA 4.0 DEED license. available on arxiv

IntelliGame in Action: Gamifying JavaScript Unit Tests - Conclusions, and References

Bitcoin NFTs, or non-fungible tokens on the Bitcoin blockchain, have emerged as a groundbreaking trend reshaping the digital asset landscape. These tokens represent ownership of digital assets and are gaining significant traction within the community. cryptocurrency In this article, let's delve into what Bitcoin NFTs are and how they are created, as well as explore the current trends surrounding this innovative technology. If you run a Web3/NFT project and need advice on marketing and promotion of your project, you can [message me](mailto: serge@x10.agency) directly. Understanding Bitcoin NFTs Bitcoin NFTs represent ownership of digital assets, such as artwork, collectibles, and in-game items. Each NFT has a unique cryptographic signature that verifies its authenticity and ownership. This means that each NFT is one-of-a-kind and cannot be replaced or replicated. This uniqueness is what underpins the value and appeal of NFTs in the digital realm. Creation of Bitcoin NFTs The creation of Bitcoin NFTs involves a process known as "Ordinal Inscriptions," where data is directly inscribed on-chain within the Bitcoin network. This method differs from Ethereum NFTs, which rely on off-chain metadata that can be altered. Ordinal Inscriptions allow for all data to be permanently recorded on the blockchain, ensuring transparency and immutability in token ownership. The process of creating a Bitcoin NFT involves several steps. First, the digital asset is created or selected, such as a piece of artwork or an in-game item. Next, the asset is tokenized by creating a unique cryptographic signature that represents ownership. This signature is then inscribed on the Bitcoin blockchain using Ordinal Inscriptions. Once the inscription is complete, the NFT is created and can be bought, sold, or traded on various online marketplaces. Trends in Bitcoin NFTs The trend of Bitcoin NFTs is rapidly evolving, with a surge in interest and adoption witnessed across the cryptocurrency space. Recent data indicates that Bitcoin NFTs have surpassed Ethereum NFTs in trading volume, signaling a shift towards the Bitcoin blockchain for NFT transactions. This is due in part to the growing popularity of Ordinals, which have made it easier and more accessible to create and trade NFTs on the Bitcoin blockchain. Some examples of Bitcoin NFTs include: : The first NFT collection ever released on Bitcoin, created in September 2016 using the Counterparty protocol. Despite their limited utility due to the constraints of Counterparty's scripting language at the time, Rare Pepes laid the foundation for subsequent Bitcoin NFT projects and highlighted the potential for digital collectibles on the Bitcoin network. Rare Pepes A metaverse project in the Bitcoin ecosystem based on an open-source standard protocol proposed on Ordinals, primarily for the primary market sale of virtual land. What sets Bitmap apart is its unique binding to blocks, creating inherent scarcity within its virtual land offerings. By leveraging Ordinals and the Bitcoin blockchain, Bitmap showcases a novel way of tokenizing virtual assets with built-in scarcity, offering a fresh perspective on digital ownership and value creation in the metaverse space. Bitmap: : A collection of NFTs on the Bitcoin blockchain, with examples like Mineral having a market cap of 379.6 BTC (approximately $19.8M). This series symbolizes the intersection of art, technology, and blockchain innovation, showcasing the potential for both creative expression and value generation. Ordinal Punks NFT series How To Make Your Own Bitcoin NFT? Creating your own NFT on the Bitcoin blockchain is a relatively straightforward process, and there are several tools and platforms available to help you get started. Here are the basic steps for creating your own NFT: 1. : The first step in creating an NFT is to choose the digital asset that you want to tokenize. This could be a piece of artwork, a collectible, an in-game item, or any other type of digital asset. Choose your digital asset 2. : Once you have chosen your digital asset, you will need to create a digital representation of it. This could be a digital file, such as a JPEG or PNG image, or it could be a 3D model or other type of digital asset. Create a digital representation 3. : Once you have created a digital representation of your asset, you can mint your NFT using a platform that supports Ordinal Inscriptions on the Bitcoin blockchain. There are several platforms available, including Raretoshi, Counterparty, and Tokenly. These platforms provide tools for creating and managing NFTs on the Bitcoin blockchain. Mint your NFT 4. : When you mint your NFT, you will need to set its metadata, which includes information about the asset, such as its name, description, and image. This metadata is stored on the blockchain along with the NFT, and it provides important context and information about the asset. Set your metadata 5. Once your NFT is minted and its metadata is set, you can list it for sale on an online marketplace that supports NFTs on the Bitcoin blockchain. Some popular marketplaces include OpenBazaar, BitPieces, and Rare Pepe Wallet. List your NFT for sale: 6. Once your NFT is listed for sale, you will need to promote it to potential buyers. This could involve sharing it on social media, creating a website or blog to showcase your NFTs, or reaching out to collectors and investors in the NFT community. Promote your NFT: Market Dynamics and Challenges While Bitcoin NFTs are gaining momentum, challenges such as scalability and accessibility remain key areas of focus for developers and users. The integration of Ordinals into the Bitcoin ecosystem has raised debates regarding block space usage and transaction costs. As more Ordinals are inscribed with additional data like images and audio clips, concerns about on-chain congestion and confirmation times have surfaced. Scalability is a major challenge for the Bitcoin blockchain, as it has a limited capacity for processing transactions. This means that as more NFTs are created and traded on the blockchain, it could become congested and slow, leading to higher transaction fees and longer confirmation times. To address this challenge, developers are exploring ways to increase the scalability of the Bitcoin blockchain, such as through the use of layer-two solutions like the Lightning Network. Accessibility is another challenge for Bitcoin NFTs, as they require a certain level of technical expertise to create and trade. This can be a barrier for entry for some users, who may not have the necessary skills or resources to participate in the NFT market. To address this challenge, developers are working on creating more user-friendly tools and platforms for creating and trading NFTs on the Bitcoin blockchain. At the end of the day, Bitcoin NFTs represent a paradigm shift in how digital assets are tokenized and traded within the cryptocurrency ecosystem. The future of Bitcoin NFTs holds immense potential for reshaping the landscape of digital ownership and value exchange. As this trend continues to evolve, it will be fascinating to witness how Bitcoin NFT conquers the new market heights. P.S. Check out some of my previous articles at HackerNoon: NFT Marketing Guide - The Most Complete and Detailed Playbook Autoglyphs: The $14.6 mln NFT Phenomenon Explained Cracking the NFT Code: The Psychology Behind Buying Digital Assets Breaking Down the Next Big Thing in NFTs: ERC-404 Spotlight Nobody Buys an NFT Just for Art DeFi Community Building: A Step-by-Step Guide for Crypto Startups

Bitcoin NFTs: Trends and How to Create It

Space travel remains the dream of many people, and Intuitive Machines' has brought us one step closer. Although it “tripped” while landing on the Moon, Odysseus has managed to communicate with its control team on Earth, beaming back information and images. Odysseus is the first private spacecraft ever to land successfully on the moon — and the first American vehicle to do so since the final Apollo mission in 1972. Odysseus lander Of course, the Moon is probably only meant to serve as a practice pad and “truck stop” in the future for more ambitious missions that will take humans to Mars and elsewhere. The Moon will hopefully be used for scientific experiments, training purposes, and as a way station for future astronauts who will spend lengthy periods of time in space. While the Odysseus was unmanned, NASA has plans to send astronauts back to the Moon through its Artemis program. Billionaire and SpaceX founder Elon Musk he believes that humans could already be making their way to the red planet in just a few years. said last year Musk an ambitious plan to establish a human colony on Mars and that he aims to transport one million people to the Red Planet, stating that “civilization only passes the single-planet Great Filter when Mars can survive even if Earth supply ships stop coming.” In the past, Musk has said he plans to have a million people on and that “a on the red planet could sustain our species if all humanity is extinguished on Earth.” recently revealed Mars by 2050 permanent colony Musk has long voiced his intent to make humanity multi-planetary, often pointing to human settlement on Mars as an insurance policy for civilization in case Earth is struck by any number of calamities. With the space race in full gear, the question of who gets what contract has become a matter of serious competition. Companies like Intuitive Machines, Musk’s SpaceX, Amazon founder Jeff Bezos’ Blue Origin, or Richard Branson’s Virgin Galactic are all competing for the eventual ability to ferry Earth’s inhabitants back and forth between our planet and whatever way station or permanent habitat we build elsewhere, whether on the Moon or Mars. SpaceX's first priority is "establishing a cargo route to Mars," in the past. Musk has said Musk is relying on the much-touted Starship to advance his plans. His early plans involved sending unmanned ships to the surface of Mars and then sending people after that once the flights were proven safe. The first missions would work to convert water and CO2 found on the Mars surface into the liquid oxygen and methane fuel that powers the Starship. By doing so, SpaceX would guarantee a long-lasting fuel source. Once refueled, the Starships would be able to make the journey back to Earth. Following the initial flights, Musk wants to bring large groups of people to Mars to build colonies and factories to make the planet self-sustaining. One of the main priorities would be to build a factory for rocket fuel, essentially turning Mars into an outpost for ever-expanding space travel. In Musk’s vision, people could go from Mars to the asteroid belts, the moons of Jupiter and Saturn, and the Kuiper belt. While in the beginning stages, people on Mars would have to live inside large protected air-tight structures on the surface or even underground, eventually, though, Musk said he envisions having the ability to warm the planet to make it habitable and allow humans to live outside without the need for supplemental oxygen. But this is where some scientists begin to see problems with Musk’s grand plan. According to experts, there just on Mars that you can mobilize and create a long-lasting atmosphere the way Musk believes he can. The other problem, even if this was possible, is keeping the CO2 contained in the atmosphere. isn't enough CO2 Musk has said he understands this issue and realizes it would take a long time. He acknowledged that this is a project that would span several generations, and even if we cannot create an atmosphere now, we could at least build the first colony on Mars, preparing the way for future generations. One of the greatest advantages of space exploration in today’s day and age is that we now have new technologies, such as artificial intelligence (AI), to expand our capabilities more than was possible when space travel first became possible. For instance, AI is to help scientists analyze data more quickly and accurately and automate operations on spacecraft. The continuous development of other planets and moons in ways never before possible. currently being used in the space industry AI is helping rovers and landers explore “SpaceX has been improving and refining how rockets operate. The company uses AI to monitor and analyze data from the rocket’s sensors and telemetry systems, allowing for better decision-making and more precise control of the rocket’s trajectory and speed. SpaceX also uses AI to automate certain aspects of the rocket landing procedure.” NASA’s Jet Propulsion Laboratory has also recently developed an that can autonomously plan a spacecraft mission based on given parameters and constraints. Machine learning algorithms are being used to identify potential water sources or mineral deposits on the planet’s surface. AI-based mission planning system Clearly, AI is essential to the future of space exploration and will increasingly be implemented across various future missions in space—from Mars colonization efforts to interplanetary travel to supporting astronaut health. We are getting closer to this vision of becoming an interplanetary species. As long as it does not hallucinate or generate false information, as some AI technologies currently do, AI will help us achieve this unique and incredible vision.

AI Will Help Mankind Colonize Mars

This paper is available on arxiv under CC 4.0 license. Authors: (1) Zhouqiang Jiang, Meetyou AI Lab, (2) Bowen Wang, Institute for Datability Science, Osaka University, (3) Tong Xiang, Meetyou AI Lab, (4) Zhaofeng Niu, Department of Computer Science, Qufu Normal University, (5) Hong Tang, Department of Information Engineering, East China Jiaotong University, (6) Guangshun Li, Department of Computer Science, Qufu Normal University, (7) Liangzhi Li, Meetyou AI Lab. Table of Links Abstract & Introduction Related Work Method Experiments Conclusion & References V. CONCLUSION In this paper, we proposed CatMAE for self-supervised video representation learning. It leverages a concatenated information channel masking strategy to address the limitations posed by cube masking and enhances the capability to capture continuous and long-term motion compared to asymmetric masking. Our experimental results demonstrate superior performance in comparison to state-of-the-art methods across both video segmentation and action recognition tasks. One distinctive feature of our training pipeline is the propagation of reconstruction information from the initial frame throughout the entire video sequence. This theoretically unlimited propagation showcases CatMAE’s potential to learn long-term video representations. Our future work focuses on extending the application of CatMAE to realworld scenarios involving embodied agents, such as robots. REFERENCES [1] K. He, X. Chen, S. Xie, Y. Li, P. Dollar, and R. Girshick, “Masked ´ autoencoders are scalable vision learners,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 16 000–16 009. [2] Z. Tong, Y. Song, J. Wang, and L. Wang, “Videomae: Masked autoencoders are data-efficient learners for self-supervised video pretraining,” Advances in neural information processing systems, vol. 35, pp. 10 078–10 093, 2022. [3] C. Feichtenhofer, Y. Li, K. He et al., “Masked autoencoders as spatiotemporal learners,” Advances in neural information processing systems, vol. 35, pp. 35 946–35 958, 2022. [4] A. Gupta, J. Wu, J. Deng, and L. Fei-Fei, “Siamese masked autoencoders,” arXiv preprint arXiv:2305.14344, 2023. [5] A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929, 2020. [6] E. H. Adelson and J. R. Bergen, “Spatiotemporal energy models for the perception of motion,” Josa a, vol. 2, no. 2, pp. 284–299, 1985. [7] J. Carreira and A. Zisserman, “Quo vadis, action recognition? a new model and the kinetics dataset,” in proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 6299–6308. [8] Z. Zhang and D. Tao, “Slow feature analysis for human action recognition,” IEEE transactions on pattern analysis and machine intelligence, vol. 34, no. 3, pp. 436–450, 2012. [9] J. Pont-Tuset, F. Perazzi, S. Caelles, P. Arbelaez, A. Sorkine-Hornung, ´ and L. Van Gool, “The 2017 davis challenge on video object segmentation,” arXiv preprint arXiv:1704.00675, 2017. [10] W. Lotter, G. Kreiman, and D. Cox, “Deep predictive coding networks for video prediction and unsupervised learning,” arXiv preprint arXiv:1605.08104, 2016. [11] M. Mathieu, C. Couprie, and Y. LeCun, “Deep multi-scale video prediction beyond mean square error,” arXiv preprint arXiv:1511.05440, 2015. [12] Q. Zhou, X. Liang, K. Gong, and L. Lin, “Adaptive temporal encoding network for video instance-level human parsing,” in Proceedings of the 26th ACM international conference on Multimedia, 2018, pp. 1527– 1535. [13] H. Jhuang, J. Gall, S. Zuffi, C. Schmid, and M. J. Black, “Towards understanding action recognition,” in Proceedings of the IEEE international conference on computer vision, 2013, pp. 3192–3199. [14] P. Vincent, H. Larochelle, Y. Bengio, and P.-A. Manzagol, “Extracting and composing robust features with denoising autoencoders,” in Proceedings of the 25th international conference on Machine learning, 2008, pp. 1096–1103. [15] P. Vincent, H. Larochelle, I. Lajoie, Y. Bengio, P.-A. Manzagol, and L. Bottou, “Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion.” Journal of machine learning research, vol. 11, no. 12, 2010. [16] D. Pathak, P. Krahenbuhl, J. Donahue, T. Darrell, and A. A. Efros, “Context encoders: Feature learning by inpainting,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2536–2544. [17] R. Zhang, P. Isola, and A. A. Efros, “Colorful image colorization,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III 14. Springer, 2016, pp. 649–666. [18] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” arXiv preprint arXiv:1810.04805, 2018. [19] H. Bao, L. Dong, S. Piao, and F. Wei, “Beit: Bert pre-training of image transformers,” arXiv preprint arXiv:2106.08254, 2021. [20] X. Dong, J. Bao, T. Zhang, D. Chen, W. Zhang, L. Yuan, D. Chen, F. Wen, N. Yu, and B. Guo, “Peco: Perceptual codebook for bert pre-training of vision transformers,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 1, 2023, pp. 552– 560. [21] C. Wei, H. Fan, S. Xie, C.-Y. Wu, A. Yuille, and C. Feichtenhofer, “Masked feature prediction for self-supervised visual pre-training,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 14 668–14 678. [22] Z. Xie, Z. Zhang, Y. Cao, Y. Lin, J. Bao, Z. Yao, Q. Dai, and H. Hu, “Simmim: A simple framework for masked image modeling,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 9653–9663. [23] J. Zhou, C. Wei, H. Wang, W. Shen, C. Xie, A. Yuille, and T. Kong, “ibot: Image bert pre-training with online tokenizer,” arXiv preprint arXiv:2111.07832, 2021. [24] M. Chen, A. Radford, R. Child, J. Wu, H. Jun, D. Luan, and I. Sutskever, “Generative pretraining from pixels,” in International conference on machine learning. PMLR, 2020, pp. 1691–1703. [25] J. Walker, C. Doersch, A. Gupta, and M. Hebert, “An uncertain future: Forecasting from static images using variational autoencoders,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part VII 14. Springer, 2016, pp. 835–851. [26] A. Gupta, S. Tian, Y. Zhang, J. Wu, R. Mart´ın-Mart´ın, and L. FeiFei, “Maskvit: Masked visual pre-training for video prediction,” arXiv preprint arXiv:2206.11894, 2022. [27] C. Vondrick, H. Pirsiavash, and A. Torralba, “Anticipating visual representations from unlabeled video,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 98– 106. [28] D. Pathak, R. Girshick, P. Dollar, T. Darrell, and B. Hariharan, ´ “Learning features by watching objects move,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 2701–2710. [29] N. Srivastava, E. Mansimov, and R. Salakhudinov, “Unsupervised learning of video representations using lstms,” in International conference on machine learning. PMLR, 2015, pp. 843–852. [30] X. Wang, A. Jabri, and A. A. Efros, “Learning correspondence from the cycle-consistency of time,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 2566–2576. [31] X. Wang and A. Gupta, “Unsupervised learning of visual representations using videos,” in Proceedings of the IEEE international conference on computer vision, 2015, pp. 2794–2802. [32] C. Vondrick, A. Shrivastava, A. Fathi, S. Guadarrama, and K. Murphy, “Tracking emerges by colorizing videos,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 391–408. [33] B. Fernando, H. Bilen, E. Gavves, and S. Gould, “Self-supervised video representation learning with odd-one-out networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 3636–3645. [34] H.-Y. Lee, J.-B. Huang, M. Singh, and M.-H. Yang, “Unsupervised representation learning by sorting sequences,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 667–676. [35] I. Misra, C. L. Zitnick, and M. Hebert, “Shuffle and learn: unsupervised learning using temporal order verification,” in Computer Vision– ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14. Springer, 2016, pp. 527–544. [36] D. Wei, J. J. Lim, A. Zisserman, and W. T. Freeman, “Learning and using the arrow of time,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 8052–8060. [37] R. Hadsell, S. Chopra, and Y. LeCun, “Dimensionality reduction by learning an invariant mapping,” in 2006 IEEE computer society conference on computer vision and pattern recognition (CVPR’06), vol. 2. IEEE, 2006, pp. 1735–1742. [38] S. Becker and G. E. Hinton, “Self-organizing neural network that discovers surfaces in random-dot stereograms,” Nature, vol. 355, no. 6356, pp. 161–163, 1992. [39] K. He, H. Fan, Y. Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9729–9738. [40] T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in International conference on machine learning. PMLR, 2020, pp. 1597–1607. [41] X. Chen and K. He, “Exploring simple siamese representation learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 15 750–15 758. [42] J.-B. Grill, F. Strub, F. Altche, C. Tallec, P. Richemond, ´ E. Buchatskaya, C. Doersch, B. Avila Pires, Z. Guo, M. Gheshlaghi Azar et al., “Bootstrap your own latent-a new approach to self-supervised learning,” Advances in neural information processing systems, vol. 33, pp. 21 271–21 284, 2020. [43] M. Caron, H. Touvron, I. Misra, H. Jegou, J. Mairal, P. Bojanowski, ´ and A. Joulin, “Emerging properties in self-supervised vision transformers,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 9650–9660. [44] M. Caron, I. Misra, J. Mairal, P. Goyal, P. Bojanowski, and A. Joulin, “Unsupervised learning of visual features by contrasting cluster assignments,” Advances in neural information processing systems, vol. 33, pp. 9912–9924, 2020. [45] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems, vol. 30, 2017. [46] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778. [47] I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101, 2017. [48] E. Hoffer, T. Ben-Nun, I. Hubara, N. Giladi, T. Hoefler, and D. Soudry, “Augment your batch: Improving generalization through instance repetition,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 8129–8138. [49] A. Jabri, A. Owens, and A. Efros, “Space-time correspondence as a contrastive random walk,” Advances in neural information processing systems, vol. 33, pp. 19 545–19 560, 2020. [50] X. Chen, H. Fan, R. Girshick, and K. He, “Improved baselines with momentum contrastive learning,” arXiv preprint arXiv:2003.04297, 2020. [51] J. Xu and X. Wang, “Rethinking self-supervised correspondence learning: A video frame-level similarity perspective,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 10 075–10 085.

Concatenated Masked Autoencoders as Spatial-Temporal Learner: Conclusion & References

For many years Nginx has been a default solution to serve as a reverse proxy for Rails applications. However, with the release of Kamal, the Rails community opened Traefik as a new reverse proxy solution. Within my 15 years of experience with Rails, I created an almost perfect configuration for Nginx that migrated through all my projects. With Traefik, I had to start from scratch. Traefik Intro Traefik is an Edge Router; it means that it’s the door to your platform and that it intercepts and routes every incoming request but does not process requests itself. When an incoming request comes to Traefik, it is getting served through the EntryPoint. Traefik can listen to multiple EntryPoints, and each EntryPoint is associated with a port and a protocol (HTTP/HTTPS, TCP, UDP). After the request is received, it is passed to the Router. Router is a rule that tells Traefik how to handle the request. It can be based on the path, the host, the headers, etc. The router will then forward the request to the appropriate service. But before a request is forwarded, it can be modified by the Middleware. Middleware is a piece of code that can modify the request or the response before it reaches the service. Middleware is the place where you can add authentication, rewrite the path, add headers, etc. To better understand how Traefik works, I provide an image from the official . Traefik documentation Compared to Nginx, the Traefik configuration is entirely different. Besides a traditional configuration file (which can be both in YAML and TOML formats), Traefik can be configured with dynamic configuration. It means that you can configure Traefik with a file, but also with a REST API, a Docker label, a Kubernetes annotation, etc. Since Kamal is a Docker orchestration tool, Traefik configuration is done with Docker labels. It means that you don’t need to write a configuration file; instead, you should provide labels to your Docker containers. Labels work dynamically, so you don’t need to reboot Traefik to apply them. At the same time, operating a Docker label can be pretty tricky, especially at the beginning. Traefik Dashboard The good news is that Traefik works pretty good with Kamal out of the box with Kamal (Kamal provides basic default configuration, which I will show later), and maybe you won’t need to change anything. However, if you want to take a look at how Traefik is configured, you may want to enable a . web dashboard To do this, open your Kamal config (usually ), and add the following lines. config/deploy.yml # config/deploy.yml

traefik:
  args:
    api.dashboard: true
  labels:
    traefik.http.routers.dashboard.servixce: api@internal
    traefik.http.routers.dashboard.rule: "Host(`traefik.onetribe.com`)"
    traefik.http.middlewares.auth.basicauth.users: user:$2y$05$us7dsDg56EJ/qojvhLvy9OEshrzoWjGSKXziBqeiFq3Ehf1pAiGSG
    traefik.http.routers.dashboard.middlewares: auth@docker If you have a separate configuration for production and staging environments, you probably will have the last two labels in your environment-specific configuration files. # config/deploy.production.yml

traefik:
  labels:
    traefik.http.routers.dashboard.rule: "Host(`traefik.onetribe.team`)"
    traefik.http.middlewares.auth.basicauth.users: user-production:$2y$05$us7dsDg56EJ/qojvhLvy9OEshrzoWjGSKXziBqeiFq3Ehf1pAiGSG

# config/deploy.staging.yml

traefik:
  labels:
    traefik.http.routers.dashboard.rule: "Host(`traefik.staging.onetribe.team`)"
    traefik.http.middlewares.auth.basicauth.users: user-staging:$2y$05$us7dsDg56EJ/qojvhLvy9OEshrzoWjGSKXziBqeiFq3Ehf1pAiGSG Let’s go line by line in file. config/deploy.yml On line enables Traefik Dashboard, as it is disabled by default. 5, api.dashboard: true On line I defined a router that will serve the dashboard. The internal Traefik API service is called ; I attached it to the router. 7, api@internal dashboard Then, on line I defined a rule that tells Traefik to serve the dashboard on the domain. Rule can be based . 9, traefik.onetribe.team on the path, the host, the headers, etc Since I don’t want to expose the dashboard to the public, I added a basic authentication middleware on lines and . Traefik supports different authentication methods: , , , etc. I used basic authentication, as it is the easiest to configure. 9 10 basic digest forward Password can be encrypted with the command, which is a , and this is really amazing how software, which is 20 years old, is still in use. To encrypt a password for a new user, run the following command. htpasswd part of the Apache toolkit htpasswd -nB user Traefik Logging Besides the visualization of the configuration, it is important to have good logging. Most of my Kamal projects use AWS CloudWatch for logging, so I configured Traefik to send logs to CloudWatch. To do this, I added the following lines to the file. config/deploy.yml logging:
  driver: awslogs
  options:
    awslogs-region: us-east-1
    awslogs-group: application
    awslogs-create-group: true
    tag: "-" This will configure logging for your application containers. In the example above, I used as a log group; you can use any name you want. The option is important; I used the container name and ID to distinguish logs from different containers. The picture below demonstrates how logs are organized in the CloudWatch. application tag As you see, besides application container logs, I also have a Traefik log stream, which includes both error and access logs (which is not the best, but it is how it is). Traefik logging can be configured by passing arguments to the Traefik container. I assume that my health check requests are pretty fast, so I set the minimum duration to 50ms to filter not important data. traefik:
  args:
    log.format: json
    accesslog: true
    accesslog.format: json
    accesslog.filters.minduration: 50ms Middlewares 101 I already talked about middlewares earlier in this article; let’s have a closer look at them. Middlewares are a piece of code that can modify the request or the response before it reaches the service. I used a basic authentication middleware to protect the Traefik dashboard. As we already know, each router has a set of settings and is also a setting that accepts a list of middlewares. Each middleware should be defined with it is own labels. middlewares traefik.http.middlewares.<middleware_1>.basicauth.users: username:<encrypted_password>
traefik.http.middlewares.<middleware_2>.ipwhitelist.sourcerange: 127.0.0.1/32
traefik.http.routers.<router_name>.middlewares: <middleware_1>, <middleware_2>, <middleware_3> : Traefik middlewares are not global. They are attached to the router, so you need to define them for each router separately. Important Kamal adds a to each service that uses Traefik. It is essential to know because if you add a middleware to a router, it will replace the default middleware. If you want to add a middleware to the default middleware, you must add it to the list of middlewares. default middleware WWW Redirect Redirecting from a www to a non-www domain is a common practice. It is important to have a single domain for SEO and security reasons. In Nginx, I used to do this with a simple directive. rewrite server {
  server_name www.onetribe.team;
  return 301 $scheme://onetribe.team$request_uri;
} To do this with Traefik, you need to configure on the router. RedirectRegex middleware servers:
  web:
    hosts:
      - <ip_address>
    labels:
      traefik.http.routers.onetribe-web.middlewares: onetribe-web-www-redirect@docker
      traefik.http.middlewares.onetribe-web-www-redirect.redirectregex.regex: ^http://www.(.*)
      traefik.http.middlewares.onetribe-web-www-redirect.redirectregex.replacement: https://$1 You probably noticed that I used schema in regex. Why not ? In all my configurations with Traefik, I do HTTPS termination on the load balancer before the request reaches Traefik. In Rails, 7.1 introduced to handle such configurations. http:// https:// config.assume_ssl option Compress Almost all Nginx configurations I saw had a directive. Compressing responses to save bandwidth and speed up the website is a good practice. Digital Ocean has an , explaining how HTTP compression works and how to configure it with Nginx. gzip excellent tutorial server {
  # ...
  gzip on;
  gzip_comp_level 2;
  gzip_min_length 1000;
  gzip_proxied any;
  gzip_types text/plain text/css application/json application/javascript text/xml application/xml application/xml+rss text/javascript;
  gzip_vary on;
  # ...
} Traefik has a that can be used to achieve the same result. Compress middleware servers:
  web:
    hosts:
      - <ip_address>
    labels:
      traefik.http.routers.onetribe-web.middlewares: onetribe-web-compress@docker
      traefik.http.middlewares.onetribe-web-compress.compress: true Full Example The example below combines two middleware defined above and a middleware that is added by default by Kamal–Retry. The Retry middleware reissues requests a given number of times to a backend server if that server does not reply. servers:
  web:
    hosts:
      - <ip_address>
    labels:
      traefik.http.routers.onetribe-web.middlewares: onetribe-web-www-redirect@docker,onetribe-web-retry@docker,onetribe-web-compress@docker
      traefik.http.routers.onetribe-web.rule: PathPrefix(`/`)
      traefik.http.services.onetribe-web.loadbalancer.server.scheme: http

      traefik.http.middlewares.onetribe-web-www-redirect.redirectregex.regex: ^http://www.(.*)
      traefik.http.middlewares.onetribe-web-www-redirect.redirectregex.replacement: https://$1

      traefik.http.middlewares.onetribe-web-compress.compress: true

      traefik.http.middlewares.onetribe-web-retry.retry.attempts: 5
      traefik.http.middlewares.onetribe-web-retry.retry.initialinterval: 500ms Rails Assets Handling Last, it is not about the Traefik but about the Rails configuration. With Nginx, I usually didn’t care about how my CSS, JS, and other static assets were handled; Nginx did this for me. All required headers (like compression that we discussed above and cache control) were added by Nginx. Traefik is not a web server; it is an edge router, so it doesn’t handle static assets. It means that you need to configure Rails to handle static assets properly. Stop, Rails static file handling is slow and not efficient. And you are right, or at least you were right. Modern versions of Puma handle static assets well. Besides, in 2024, we usually use cloud storage like AWS S3 to handle user uploads, so Rails will only serve CSS and JS files, which is not a big deal for Puma. I ended with something like this in my file. config/environments/production.rb # config/environments/production.rb
  # ...

  config.public_file_server.enabled = true
  config.public_file_server.headers = {
    "Cache-Control" => "public, max-age=#{30.days.to_i}, must-revalidate"
  } header is important; it tells the browser how long it should cache the file. In this example, I set it to 30 days. Must-revalidate tells the browser to revalidate the file after the cache expires. Cache-Control With such settings, Rails will serve static files only once for each client. If you think this is not enough, you can use a CDN like Cloudflare or AWS CloudFront to cache responses from Rails and dramatically reduce the load on your server. Conclusion As it usually happens with new technologies, it takes time to get used to them. Traefik is not an exception. It took me a while to understand how it works and how to configure it. At the same time, I see the benefits of using Traefik over Nginx with Rails – the main one is, of course, support of Docker out of the box. I believe this was the main reason why Basecamp decided to go with Traefik. Almost a year has passed since I switched our first project to Traefik, and I’m happy with the result. I hope you will find this article useful. Feel free to ask questions in the comments below.

Stories, News & Mentions about MINDTERA

Mindtera’s Stories on HackerNoon