deeptraffic

Startups of The Year 2023 Winners: Europe

Let's learn about via these 60 free stories. They are ordered by most time reading created on HackerNoon. Visit the to find the most read stories about any technology. Compliance /Learn Repo 1. Promoting Reliable Decentralized Finance: An Interview with 2 Experts Ever since Bitcoin first came into play in 2008, the cryptocurrency space has never been deprived of concerns. 2. Bye-bye Google Fonts: German Court Rules That Google Fonts Is Not in Compliance with GDPR Google Fonts is a data privacy concern. Learn why you should care. 3. Astra Protocol Brings Decentralized Compliance to Sandbox's Metaverse by Setting up Compliance Hub Astra Protocol, a cutting-edge legal and compliance layer that provides unparalleled security to public blockchains, announced the launch of a Compliance Hub. 4. ‘It’s Never Too Late To Learn a New Language’: Yair Solow, CEO of Centraleyes Startup interview with Yair Solow, founder and CEO of Centraleyes. Learn more about the company and different solutions. 5. Blockchain Regulations Must Empower Those Outside the Traditional Banking System Banking and finance instruments have slowly evolved from being exclusively accessible to the fortunate few to becoming widely accessible to the masses, who are now able to transact with relative ease and accumulate wealth. 6. Worried About Your Crypto's Custody? Outsource Everything Except Key Management A year into my stint at the Bank of France, I was still struggling to understand why they employ cryptographers. One day, I saw them huddled around a white board for a few hours and struck up a conversation. They walked me through their highly sophisticated key management solution that included everything from certificate signing mechanisms to deep traffic inspection contraptions. Baffled by the complexity of it all, I resorted to my go-to cryptographer question - "So, where are the keys?". The key storage, as it turned out, was outsourced to "a cool startup here in Paris.” 7. The Pros And Cons Of Outsourcing Your Cybersecurity Compliance Whether they know it or not, any business with an internet-exposed edge has to take cybersecurity seriously. 8. Why We Need Regulatory Compliance in Web3 A conversation with Sakhib Waseem of Astra Protocol on why regulatory compliance is both needed and a positive development for the Web3 space as a whole. 9. How to Build a DevOps Culture Within Engineering Teams 10. Report: 10% of Crypto Exchanges have 'Good' Cybersecurity in Place 11. Conquer GDPR Article 32: Locking Down Data Segregation and Identity-Based Access Understand the importance and tools of implementation for data segregation through access control GDPR compliance to address Article 32. 12. Faster is Safer: How Software Physics Beats Human Psychology The idea that we make software safer by releasing it slower is an illusion. Here's why. 13. How Cloud Provider Compliance Affects Your Application Compliance Cloud is evolving fast and already becoming a significant part of the software ecosystems. With this growth, becoming compliant on standards such as ISO, PCI-DSS, HIPPA, builds confidence for you to trust the cloud in terms of security. 14. A Guide to HIPAA Compliance in 2021 & Its Benefits Following HIPAA laws provides a higher level of safety for your corporation and its leaders. 15. The Massive Uber Hack: Technical and Legal Implications In this article, I analyze the technical and legal implications of post-data breach, the risks & possible exonerations in the eyes of extent laws & regulations. 16. Concordium Blockchain valued at $4.45 Billion in its Last Private Sale is now Live Concordium is a public, layer-1, proof-of-stake blockchain with a unique ID layer at the protocol level to assist with regulatory compliance. 17. 3 Solutions for Sharing and Managing Your Zoom Recordings Recorded a Zoom meeting, now what? How do I efficiently share it with people around my organization? How do I restrict access to such a recording? I want my marketing team to only view this recording so how do I segregate access? 18. Providing Proof in an Untrusting World Dragonchain's hybrid platform provides blockchain-based measurable proof  for selective transparency, accountability, and regulatory compliance. 19. Compliance is Not a Guarantee Against Data Breach When it comes to combating the ever-growing threat posed by data breaches, most organizations believe that compliance is key. With cybercriminals targeting governments, small businesses, and other essential infrastructure- the need to minimize the impact of these attacks is at an all-time high. 20. 10 Factors to Prepare for CCPA (California Consumer Privacy Act) Be ready for CCPA with these top 10 critical factors 21. Readable Legal Documents Are No Longer a "Nice-to-Have" By writing your company’s privacy policy in a clear and structured way, you influence how privacy policies are written and designed in huge organizations. 22. Cloud Compliance And Regulation Resources Customers and Cloud Service Provider (CSP) share the responsibility of security and compliance. Thus, the organization would have the freedom to have architect their security and compliance needs, according to the services they utilize from the CSP and the services they intend to achieve. CSP has the responsibility to provide services securely and to provide physical security of the cloud. If, however, a customer opts for Software-as-a-service, then the CSP provides standard compliance. Still, the organization has to check if it meets its regulations and compliance levels to strive to achieve. All Cloud services (such ad different forms of databases) are not created equal. Policies and procedures should be agreed upon between CSP and client for all security requirements and operations responsibility. 23. How to Optimize Your Salesforce Security Companies implementing Salesforce’s internal security solutions may find their customizations limit visibility into the risks that can impact other processes. 24. How Website Accessibility Affects Your Tech Brand’s Reputation and Success If your website isn't easily accessible by readers with disabilities, you aren't only losing revenue opportunities, but you’re at risk of legal problems. 25. Data Leak Prevention with Databunker We hear the horror stories of data leaks almost every day. Some require a sophisticated attack to gain shell access to the servers. Some happen over the web using the SQL injections in the web apps. As a result of this attack, your customer data can be dumped and personal data leaked. 26. How does IT Process Automation Help in Improving Business Productivity Discover how IT process automation help in improving business productivity. 27. 8 Cloud Compliance Frameworks and Why Cloud-based Organizations Need Them Cloud compliance frameworks help an organization to keep its database secure and private. In this article, you will eight such frameworks that can guide you. 28. Your CI System is a Terrible Compliance System of Record - Here are 5 Reasons Why Jenkins, GitLab, Travis, Circle CI are great for DevOps but not so great for your compliance records - here's why. 29. Establish Your Data Team Early Using These Tools and Techniques Start building your data strategy team and architecture as soon as you can. A one-person data team can make a massive difference to your success as you grow. 30. The deal requires Twitter to be compliant with Musk's request. But not with "past practice" Twitter v. Elon Musk Court Filing is part of our Legal PDF series. Part 10 of 31.FACTUAL ALLEGATIONS-The Final, Agreed-upon Term-Ordinary Course Covenant 31. New Data Privacy Laws Are an Opportunity, Not a Threat The New York Times declared the 2010s as “The Decade Tech Lost Its Way.” And it’s easy to agree when you look back at the Cambridge Analytica scandal, tech companies who consistently got off easy after privacy violations and the rise of sweeping new regulations to protect personal data. 32. How a Data Privacy API Abstracts and Simplifies Privacy and Compliance Data privacy and compliance are hard! Let a data privacy API help. 33. SOC 2 for Startups: Save Time and Money In The Long Run There is no better time for startups to undergo a SOC 2 Examination. See the top reasons how a SOC 2 audit can boost your startup. 34. FTC Mandates for Direct-Selling and Ecommerce Industries Ensuring legal compliance is crucial for direct selling companies to avoid legal issues and maintain the trust of their customers and distributors. 35. An Overview of Legal Compliance as a Service How to automate compliance with legal software. 36. Multi-Cloud Environment and Hurdles of Incorporating IAM Understand the aspects of deploying IAM in multi-cloud environments and how businesses can ensure maximum security and privacy while moving to the cloud. 37. How to Ensure your SaaS is Compliant - For Developers This article reviews what kinds of compliance certifications and regulations exist, what they mean for developers, and how to begin the certification process. 38. "Hackers Need To Get Lucky Only Once" - Excerpts on Fighting Crypto Hacks, Theft and Fraud What do trust & safety leaders expect in 2022 with the rise of blockchain and crypto? Find out why next year is the year for compliance, regtech, trust & safety 39. How Compliance Management System Tools Are Changing the Compliance Landscape Compliance management system tools are changing the compliance landscape. 40. 5 Things You Need to Know About CCPA Compliance The California Consumer Privacy Act (CCPA) is meant for those companies to comply with which have customers in the Golden State. The statute “represents a shift in perspective” for data. The state legislature has certain aspects that must be taken seriously by compliance practitioners if their companies are to be in compliance with it. 41. CCPA: Compliance Issue and Major Fraud Risk The California Consumer Privacy Act (CCPA) signals a major step in online privacy rights for California consumers and will likely make its way onto the national scene in the near future. In fact, research shows that 80 percent of businesses predict that CCPA will become national law at some point in the next five years. 42. How a Data Privacy Vault Protects PHI for Healthtech Users With healthcare apps and wearables that gather customer PHI, healthtech companies need a data privacy vault to ensure HIPAA compliance and build customer trust. 43. International Tech Sanctions: A Summary for Startup Founders In this article I'll give you a quick overview of the concept of international sanctions, how they work and how it may impact a FinTech company, especially from the standpoint of technical implementation and tools. 44. 10 Things I Did To Increase CloudTrail Logs Security It's crucial to safeguard CloudTrail logs and enhance the security of them, to meet compliance regulatory requirements and internal business needs. 45. Crypto's Value Is Not Relevant To Investigators, Criminals Are The UN estimates $2 trillion annually is involved in money laundering, and a small but significant percentage is done with crypto. Despite the crypto winter... 46. An In-depth Look at Shadow IT How should IT departments respond to shadow IT? Is it realistic, or even practical, to track 100% of all IT resources? Should there be a more managed approach? 47. 3 Ways to Share and Manage Zoom Recordings Recorded a Zoom meeting, now what? How do I efficiently share it with people around my organization? How do I restrict access to such a recording? I want my marketing team to only view this recording so how do I segregate access? 48. People, Not Systems, are the Real Vulnerabilities in the Workplace Why companies need to reevaluate cybersecurity to reduce risks within the flow of work. Remote work and SaaS apps make securing people more important than ever. 49. Why Websites Need To Be ADA Compliant Did you know when a company website is not ADA Compliant, a lawsuit can be filed? 50. How to Build a Compliance Culture by Improving Employee Engagement in Your Company Creating a culture of compliance means that the organization understands that compliance is critical. 51. Understanding The Importance of SOC2 Compliance for Data Centers Businesses rely on Data Center SOC2 Audit Reports for critical business decisions concerning outsourcing services. It helps customers build a sense of trust. 52. Ethereum High Transaction Fees Discourages New Participants From Entering the Ecosystem This article talks about blockchain, FATF and NFTs. 53. 10 Important 409a Valuation Compliances to Know In this article, we will discuss some of the important considerations in the 409A valuation in order to comply with the law. 54. What Makes Zero Trust Better And Different From Traditional Security Traditional vs zero trust? Learn how zero trust outperforms the traditional model by delivering improved security, flexibility and reduced complexity. 55. LCX Announces Venture Program To Invest Up To $2.5M In Compliant Token Sales At the core of LCX’s infrastructure is their native LCX Token, aiming to elevate the platform to greater heights. 56. 5 Tips for Integrating Security into Development: Part 2 In this second part of our series, we explain five additional tips to help you code more securely, from leveraging existing frameworks to protecting data. 57. Regulatory compliance needs to become mandatory for all cryptocurrency exchanges Regulatory compliance needs to become mandatory for all cryptocurrency exchanges 58. How Novel Technology Boosts Compliance in Pharma Compliance programs aim to promote transparency, safety, and efficacy in the industry. However, achieving compliance in pharma can be complex and costly. 59. Newsflash - "People with AWS Console Access Tend to Change Things" This blog post is a written transcript of the FOSDEM Talk: “Infrastructure drifts aren’t like Pokemon, you can’t catch ’em all”, by Stephane Jourdan – CTO and founder 60. Building a MongoDB NoSQL E-Commerce Data Model MongoDB supports ACID transactions, and its cloud offering is PCI DSS compliant. It's one of the best NoSQL options for e-commerce. Thank you for checking out the 60 most read stories about Compliance on HackerNoon. to find the most read stories about any technology. Visit the /Learn Repo

60 Stories To Learn About Compliance

A year into my stint at the Bank of France, I was still struggling to understand why they employ cryptographers. One day, I saw them huddled around a white board for a few hours and struck up a conversation. They walked me through their highly sophisticated key management solution that included everything from certificate signing mechanisms to deep traffic inspection contraptions. Baffled by the complexity of it all, I resorted to my go-to cryptographer question - "So, where are the keys?". The key storage, as it turned out, was outsourced to "a cool startup here in Paris.” There are very few reasons to outsource the security risks associated with storing and managing your encryption keys. However, this what often happens at companies dealing with digital asset custody. These companies overgrow with analytics, support, marketing, and administrative personnel, but end up relying on BitGo and others to sign their transactions. Keep your key storage in house. BitGo as much as you do. Snowden taught us the importance of end-to-end encryption, but we still failed to understand the fundamental problem with outsourcing encryption. We've put too much trust in Google employees who did not care about their client’s privacy enough to . Crypto custody is hard. will never care about your keys encrypt traffic between their data centers Encryption was being worked on prior to Snowden but it didn't seem like a high priority because there was no evidence it would achieve anything useful, and it rquires a lot of resources. Google Cloud Exploitation - NSA presentation . Cryptographers are as good at implementing protocols as interior designers are at laying brick. You also can't hire astrophysics PhDs hoping that they'll figure it out on the job. They will sit there for a year reading articles and dreaming up . Neat idea, until or Vitalik decides to fork Ethereum on Friday evening. CS degrees are no guarantee of expertise time locks for your cold wallet Pearl Harbor happens on Sunday have plagued existing crypto custody providers and stopped others from joining the market. Some are straight from spy novels and TV shows. Others will likely make their way to DEFCON slides in the coming years. Amateur key management setups . OKEx cold wallets cannot survive a single signature loss . Gemini founders like traveling around the country with pieces of paper to address Armageddon risks . Bitmex execs personally sign transactions every day . Fishing rod key storage system is a thing Swiss mountains don't add an extra security layer. Airgapping cold wallets without proper access and emergency procedures is useless. IP address attribution and SMS as a second factor authentication are fundamentally broken. Hardware wallets aren't any more secure than a regular smartphone, but attract way more attention to their users. exists. So does . A few basic things to remember. Rubber-hose cryptanalysis deniable encryption What you really need is a team of highly experienced software and infrastructure engineers led by security professionals, preferably the ones that haven't . Getting them won't be easy. The post-Snowden IT security job market has been vacuumed by armies of corporate and government recruiters. Big IT companies and three-letter agencies catch talented kids well before they graduate college. Anyone who doesn't have the time or expertise to grow talent internally is mostly out of luck. Talk to your local key management professional. lost all their data at their last place of employment There are still plenty of people who are unable or unwilling to pass security clearances or pee in the cup. You can also pick up whole teams by looking outside the crypto space. Unlike what you've been told, blockchain is not magic. It's a decentralized system like many others, and we have plenty of hustling security-oriented startups staffed with talented network, database, and systems engineers. Think outside the box. Anything and anyone who doesn't need to have access to your keys can and should be outsourced. Establish relationships and offload your headaches to professionals in their respective areas. Unless you plan to , you will need to put KYC and AML programs in place. Regulators will ask you to watch hundreds of metrics and produce dozens of reports every day, but they will not say which ones upfront. You are better off letting someone else argue with them about the methodology of . Your IT department will thank you for it. Open up and focus on fundamentals. bribe officials with coconuts calculating the price of a 51% attack . While the goal of a compliance team is to produce good-looking reports for the regulators, an auditor is there to show you weak spots in your setup. Hire an outside Red Team and give them sizeable bonuses for discovering vulnerabilities. Security by obscurity only works to a certain point - publish your security designs and . Create an and let people gradually poke holes in your systems. and the ability to laugh at yourself go a long way in saving you public humiliation. Separate auditing and compliance openly talk about your failures open bug bounty program Hacker t-shirts . Any asset custody is about reputation and companies that shift their security risks to others will not last long. When their assets under management reach critical mass, many of their security mechanisms will fail at once. When someone robs a bank, getting to the vault is only the beginning. Getting away with the loot and spending it is an equal problem. There's always a chance of catching the thief and getting the money back. Not so much in crypto. Bitcoin money laundering services cost 1-4% of the total sum. Hackers will start looking into yachts the second they get to your private keys. Remember, it's a marathon Digital currency money laundering services offered at a Russian darknet marketplace. Illustration by Mihailo Tatic

Worried About Your Crypto's Custody? Outsource Everything Except Key Management

Today we are talking marketing and online business development with one of the leading experts in the field, Marco Calamassi, who started up in this industry as far back as 2012, when Social Media (Facebook) was just conceived and barely launched and now is one of the very few Marketing Experts verified on both Facebook and Instagram in the world. Bare with us because a lot of data and valuable insights will be shared and no tip will be spared. Analytics attachments included. We tried to avoid the typical trivial and superficial questions he gets asked during interviews, and instead dive deeper into technicalities and insights, it’s not everyday that you get a chance to absorb invaluable notions that will easily cost thousands of dollars in consultation fees. But first, just a bit of a background and backstory of his. Marco Calamassi, born March 20th 1986 in Italy, got his feet in the online world back when he was 10-11, in the late 90s. At that time, the Internet was just starting to get popular and online commerce was “the new thing”. Ebay quickly took over the online shopping market and the late 90s economic frenzy was just the gasoline on the fire. Marco, from his early-tech PC, jumped on the Ebay hype that was building up across the masses and quickly identified a profitable business model. He imported containers of sneakers and running shoes from China and flipped them on Ebay for 3-5x markups. From that initial spark of entrepreneurship, Marco Calamassi has never let it go. He evolved his expertise as the market and digital space evolved along with new platforms such as Amazon, Instagram and Youtube and new techniques such as Email Marketing, Social Media Management and Paid Advertising. In 2012 Marco Calamassi put his knowledge to use in helping his family business, which is Made In Italy fashion, art and decor, expand to America with a Pop Up art gallery and a Fashion boutique. With nationwide retail trend starting to show signs of a worrisome decline, Marco foresaw the opportunity of a fast online transition that brought him to become a prominent Digital Marketer with outstanding results for his clients, to be one of the few marketing personalities officially verified on socials, to collaborate and consult in the Movie & Entertainment industry and be constantly featured on magazines and publications. His book “Shaping The Reality Of 2050”, written just for friends/family and colleagues in 2018 and then published in 2020, has been considered a unique and insightful outlook at the digital society of the future where Web Personalities will be the new Rockstars, Online Entrepreneurs the new Tycoons and Marketers & Developers the new Philosophers and Architects. It has been cited as a “must-read” book on Dzone.com, one of the most prominent technology websites with 5m monthly visitors. He consulted for two movies and a tv series with the role of Marketing Consultant and Digital Marketing Expert. His profile and credits can be seen on IMDB.com, a website with 450m monthly visitors. Now let’s dive right into the technical stuff and understand how and why Marco Calamassi is in the top 0.1% of marketers available on the market. Let’s talk advertising and lead generation now. We asked him to provide some analytics and data from his past campaigns and here some considerations. The average cost-per-lead in verticals such as Marketing is around $5-8 for entry level offers and up to $30 for high ticket offers. Marco Calamassi has shown us numerous campaigns that generated marketing leads for less than $3. Most of them in the $1-2 cost range. Even hundreds below a dollar. And we are talking about American audience (Tier1 targeting). Let’s go even more technical and evaluate some KPI (aka Key Performance Indicator) or metrics. The holy-grail metrics in Lead Gen and Paid Advertising are CPC and CTR, respectively Cost-Per-Click and Click-Through-Rate. CPC indicates the cost we pay to obtain 1 click. CTR measures the percentage of people that interact with our Ad out of the people that are shown the Ad. Simply put, CPC gives us an idea of how much money we are spending to land visitors to our offer and CTR measures how attractive and relatable is our Ad to that target audience. Average results, in the US, are $0.70-$1.20 CPC and 2-3% CTR. Marco provided campaigns data showing CTR of up to 12%, with averages of 8-9%. And CPC hovering around $0.10. These metrics are just outstanding. One of the most sought-after metrics for any marketer on heart, is the Relevance Score. This “recognition” KPI has been recently dismissed by Facebook because it was generating too much controversy and fights between marketers and agencies trying to catch a high RS more than anything else. The RS (Relevance Score) is a rating that goes from 0 to 10 and it measures, in simple words, how good is your Ad compared to all the other Ads in the marketplace at that given time. It highlights how effective that Ad is compared to its competitors, in the auction (Paid Ads are built upon a bid-measured system where a 24/7 auction takes place and marketers “bid” on client-acquisition costs against all the other marketers). It’s said that a Relevance Score of 6 put you above the average and should yield you positive results. A RS of 7-8 put you in the top 20% of performances. Relevance Score of 9 is the holy grail while a 10 is as rare as a Unicorn. Marco’s campaigns show an incredibleamount of 8s, 9s and 10s. One of the most popular advertising objectives is Engagement. Whether you need to grow your Facebook Business page or drive views to a video or spark a conversation with your ideal customer, Engagement is what you need. An engagement campaign to perform requires a deep understanding of your audience and who your target customer is. To know their mindset and psychology inside out. If this analysis has been done correctly and the targeting is on point as well, what is left is a powerful copywriting and a perfect medium to transmit your message over to your audience. By medium we mean the right image or video. If all this has been nailed down, the average CPE (Cost-Per-Engagement) is around $0.10 in the US and $0.01 Worldwide. Marco sent us an Engagement campaign he ran 2 years ago with the aim of boosting a music video for an artist. We couldn’t believe the screenshot. Marco told us that this campaign and screenshot went viral on Facebook groups of marketers such as “Facebook Ad Buyers” and “Facebook Ads Rockstars”. These are communities where elite marketers share tips, insights and the latest procedures. As you can see on the screenshot, this campaign was getting Engagement for 1/100th of a cent (= $0.0001). It’s even hard to write it and comprehend it. In simple words, it was yielding results for free. Mark Zuckerberg wasn’t happy, for sure. That campaign reached 818,794 people with a $67 spend. Just phenomenal. Now we want to transition into Conversion ads, those used to make a sale on the spot, whether is a physical product or info-product or a service. Marco has numerous successful campaigns but for the sake of time and to not annoy our readers we will present one case study only. For this purpose we are going to let Marco Calamassi speaks for himself cause these insights are too hard even to just transcript: “One of the most important parts of being successful with advertising when it comes to “cold traffic” (= people that never saw you, your brand or product before) is putting smart work into building out creatives (=the visual aspect of an Ad). Sure it is important to have good targeting, understanding the algorithm and manual bidding, but make sure your creatives are the killer of the game. Together with the clients, I always build out creatives that directly speak to the customer’s mind and emotion, depending on which step they are in the “purchase journey”. Customers in different stages (they see your Ad, they search you on Google, they email you, they message you on Instagram, they buy from you….all different stages) have different needs and you need to address those. Let me give you a quick story that serves as an example: You are selling an Antivirus Software and you start a conversation with somebody that saw your Ad and needs an Antivirus right now, he therefore asks you ‘I think I have a Virus in my PC, can your software remove it?’ And you keep bombarding him with the same info from the Ad which was: ‘Our Antivirus is the best one in the world’ and so he asks you again ‘Can your software remove this?’ and you keep telling him ‘We have over 1000 5-Star reviews, check it out now.’ Then he leaves angrily and buys from somebody else. formal attitude, you don’t go for a kiss straight away. Same for a customer that bought from you multiple times, you don’t send him a formal email, he is “family” by now and you must treat him and communicate with him in a much deeper and informal way. So, in conclusion ,make sure you treat every customer differently depending on which step they are at. Here is a little road map: – Cold Traffic – Point out the problem your product solves. Try to give them the feeling that they have the problem and offer them the solution, packaged as a great deal. – Warm Traffic – Try to showcase your product even more. Introduce it as the ‘Hero to all of their problems’. Connect on an emotional level. Read their mind. Offer what their mind is seeking. – Hot Traffic – Show them some social proof, reviews of your product, how people use, testimonials so they can relate and think “many bought before me, OK This conversation makes no sense and it’s the same with the ads you show to people. On every step of their journey, you need to communicate with them in different ways. It’s like a date. First time you engage with your date, you highlight those you think are your qualities and use a , it’s the last push I needed” A trick I often use to give that final push is using “flash sales”, with those I have been able to generate 10x ROAS (Return On Ad Spend) even with just warm traffic. Here an example of a campaign that yielded a 6.5X ROAS in just a month. One crucial thing I want to point out is how important CR is. CR (= Conversion Rate) is the metric that, mostly in eCommerce, shows what percentage of visitors translated into buyers. Over the years I have seen many marketers showcase big Revenue numbers without showing the Ad Spend. CR is the ultimate performance indicator. With a high CR you can literally print money out of this air = generate big revenue with little spending. Average CR in eCommerce, on websites built on Shopify (the industry standard) is around 2.5-3%. On sales funnels that can go up to 5-6%. Thanks Marco, it’s always interesting to learn new things even in a realm that look so difficult and counter-intuitive. As your CV and track record shows, Social Media growth, management and content creation are other areas of interest and expertise of yours. Thanks for sending some results for that particular sector as well. These are 2 influencers clients of yours you helped build their social following. What strategies did you use ? We all know how difficult if not impossible is to grow on Instagram these days. “Yes and because of that difficulty, Instagram growth is the most demanded service in the marketing space (he smiles). When social media are in their initial stage, right after inception, the OR (=Organic Reach) is high, it’s like a carburetor that is fully open and lets the whole flow of gasoline to reach the engine. Over time, that flow is limited. Same with the Organic Reach. These platforms want to attract as many users as possible to grow their market share and boost their numbers to obtain funding and investments and so on. Once that is done, they lower that reach because the next stage in their business plan is to develop their own built-in Advertising offer. Translated, you cannot grow anymore unless you spend in Advertising. Now, for any issue there is a fix. For any roadblock, there is a workaround. Without going too much into details, in the last 6 months I have grown my clients following through Giveaways. They are promotional posts hosted on big/huge celebrities’ accounts where big prizes such as cash or Iphones or even cars are offered to the winners. To obtain an entry in the giveaway the users must Like the post, FOLLOW those followed by the celebrity (= the giveaway clients) and tag as many friends as possible. It’s perfectly legal and it yields incredibly fast results. For a stable, consistent growth instead there is no other option than qualitative CONTENT CREATION. Which is, like for Ads creatives, simply building images or videos that speak to your audience and connect on an emotional level. And the goal is for the content to go viral and be reposted, shared, tagged generating a viral effect.” This has been a long and detailed interview with a leading Marketing expert. It’s a rarity to have such availability and depth of sharing. We wanted to offer our readers a more profound and analytical overview of what advanced and effective Marketing involves these days. With the recent outbreak of Covid-19, retail and brick and mortar took a hit that will unlikely be gone anytime soon. We have seen a huge influx of businesses and companies into the digital realm, whether they were trying to completely transition from offline to online or simply open up their “web” department. Marketing has never been so popular and in need but navigating the professional offers is hard. Marco Calamassi has an outstanding track record and 360 degrees overview of all the latest and more effective trends and techniques in the space. Hoping you found this interesting, be on the lookout for future experts and their technical expertise to be showcased.

Learning Advanced Marketing with Digital Expert Marco Calamassi

My 2018 Self Driving and Flying - Learning Path Blog Series. , connect with me on You can find me on Twitter @bhutanisanyam1 Linkedin here This is Part 3 of the series. Here is the Part 2, go check it out to know what has been going on in my Garage! 1 Month down the Highway with My Self Driving Car Path and I’m loving the ride. And I have some exciting News to Share. Learning But First, here’s what I’ve been upto in my Garage: Completed the and got a High Score of 70Mph in DeepTraffic. I’m looking forward to the release of SegFuse to try more exciting stuff. MIT SDC 2018 Course [Failure] Fried the Camera on My Mini SDC Project. Fair Warning: Be very careful when working with Bare Circuit Electronics. Luckily, my New Camera has just arrived as Today, the Pi Board was undamaged and I can get back to it. RasPi Completed the Projects Lane Finding and Traffic Sign Classifier. A detailed post on these once these have been reviewed by Udacity. Practised LSTM implementations (End-to-End) with Keras (Link TBD). Learnt ROS Basics. [Announcement]: I will soon release a Medium and YouTube Tutorial series on working with Self Driving Cars using ROS (with Gazebo Sim). Here’s the First (Unedited) Video from the Series: _Edit description_drive.google.com Sanyam Bhutani_ROS for SDC Part 1.mov I’ve explored a Few Simulators for Playing with my SDC Ideas: - - - - - - A more detailed Post on these soon once I have a firm understanding of all the Simulators. Right Now, I like ETS2 and Carla a lot. But I have a lot more tinkering to do. CarlaSim Autoware AirSim TrueVision RightHook EuroTruckSim 2 Google India Scholarship I have always wanted to work on Deep ideas that would help people at a massive scale. Learning Developing Android/iOS applications would be the ideal path to this. Thanks to Udacity and for granting me the Scholarship to Android App Dev. Google I certainly hope to release a few ideas of mine (For free of course, by Q3–2018). Changing Lanes to Take a Flight Given my belief in Sebestian Thrun’s philosophy of Dreaming Big and chasing ideas fearlessly, I was really looking forward to the Flying Car Nanodegree. And when I received an acceptance mail from Udacity, On the 14th of Feb - I decided to change lanes and Shift Gears for a Flight. What would you expect a Geek to do on Valentine’s? I decided to invest my Savings for a GPU-Rig to learn about Flying Cars. Fly Now, Train Models Later. (Meanwhile, AWS to The Rescue) I’m definitely up for a Challenge since I’m studying about Flying Cars, Self Driving Cars and Learning from the Scholarship and I’m a ‘Full Time University Student, Part Time Deep Learning Student’. But I certainly will continue sharing in this series, what I’ve learnt via the Nanodegrees. And Share my Path to Driving and Flying Autonomously. What are you geeks upto on Valentine’s? Re:Coming up Soon (Links TBD): PyTorch Series ROS Series SDC Projects Series Flying Car Series (Detailed Project Descriptions, If I’m able to- A YouTube Series) Here’s a Long Term Dream Concept of Mine: Source: https://i.ytimg.com/vi/DBz_8Cif-YY/maxresdefault.jpg , connect with me on You can find me on Twitter @bhutanisanyam1 Linkedin here Subscribe to my Newsletter for a Weekly curated list of Deep learning, Computer Vision Articles

A Self Driving and Flying (New) Year #3

, connect with me on and are two articles on my You can find me on Twitter @bhutanisanyam1 Linkedin here Here Here Path to Self Driving Cars ing You can find the Markdown File Here You can find the Lecture 1 Notes here Lecture 2 Notes can be found here Lecture 3 Notes can be found here Lecture 4 Notes can be found here These are the notes for the Course (2018), Taught by . Lecture 4 MIT 6.S094: Deep Learning for Self-Driving Cars Lex Fridman All Images are from the Lecture Slides. Applying DL to understanding Sense of Human Beings Focus on . Computer Vision How can we use CV to extract useful information from Videos (in Context of Cars) Using CV, DL to create systems that operate in the real world. Deep Learning for Human Sensing: Requirements (Ordered according to importance): Data:Enormous amounts of real data is required. Data collection is the hardest and most important part. Semi-Supervised:The raw data needs to reduced to meaningful representative cases, raw data needs to be annotated. We need to collect data and use semi-supervised techniques to find pieces of data that can be used to train our networks Efficient Annotation:Good annotation allows good performance. Annotation techniques for different scenarios are completely different. ex: Annotation tools for glance classification Vs Annotation for Body Pose Estimation Vs Image Pixel level labelling for SegFuse Hardware:Large amount of data needs large scale distributed compute and storage. Algorithms:We want algorithms that can self calibrate, allowing generalisation. Temporal Dynamics:Current algorithms are majorly image based Vs Temporal/Sequence based. Takeaway: Data Collection, cleaning is more important than algorithms. Human Imperfections Distracted Driving:3,179 people were killed and 431k+injured in crashes involving distracted driving during 2014. Eyes off the road:5 seconds is the average time, your eyes are off the road while texting. Drunk Driving:Accountable for 31% of the traffic fatalities of 2014. Drugged Driving:23% of night drivers are drugged drivers (2014). Drowsy Driving:3% of all traffic fatalities involved a drowsy driver. Given these flaws, and the Two Paths to an Autonomous Future (Human Centred Vs Full Autonomy) discussed in Lecture 1: Is the Human Centred idea a bad idea? Humans might tend to ‘over-trust; the System. MIT-AVT Naturalistic Driving Dataset Data Collection: Two+ One Cameras. : Capturing HD Video of the Face for Glance Recognition and Estimating Cognitive Load. Camera 1 : (FishEye) Estimating Body Pose-Hands on Wheel, Activity Recognition. Camera 2 : Recording scenario outside for Full Scene Segmentation. Camera 3 The Data collected provides an insight of Human Behaviour. Deploying Autonomy. Design of Algorithms for Training the Deep Neural Nets for Perception Tasks. Safety Vs Preference for Autopilot? The Dataset shows that the Physical Engagement remains the same with/without the Autopilot on. So, Autopilots allow being physically engaged. But the Systems aren’t over-trusted by the Driver Pedestrian Detection Challenges: Different appearances: Inter class variations. Different Articulations. Occlusion accessories. Pedestrians occluding each other. Solutions:The need is to extract features from raw pixels. Sliding Image: Haar Cascade. HOG. CNN. More Intelligent netoworks: Fast R-CNN. Mask RCNN. VoxelNet. These networks generate the candidates to be considered instead of a sliding window approach, providing a subset to be considered. Using a CNN classifier to detect if there is an object of interest present. Use Non-Maximum Suppression to remove overlapping bounding boxes. Data (from different intersections): 10 hours recorded every day. Aprrox 12,000 Pedestrians Crossing. 21M+ samples of feature vector. RCNN does a bounding box detection of Pedestrians. Body Pose Estimation Includes: Finding Joints in the image. The landmark points in the image. Why is it important? To determine the alignment of the driver. Note: The general airbags are deployed assuming the Driver is facing towards the front. With Increased automation, this assumption might fail. Sequential Detection Approach Detect the Hands and then detecting in steps, the shoulders and so on. Traditional Method DeepPose Holistic View: Powerful, successful for Muti-person, multi pose detection. Performing a regression of Detecting parts from the complete image individually rather than a sequential detection. Later, it stitches the detected joints together. Allows detection of varying poses, and joints that aren’t visible. Cascade of Pose Regressors: CNN that take in a raw image and produce a X-Y position of an estimate of each joint. Every estimation zooms in and produces repeated finer detection and estimation of joints. Part Detection: We can use this approach to detect parts in a picture with multiple people. First, the body parts are detected without doing individual person detection first. Next connect them together. Through Bi-Partitite matching, stitch the different people together. This is the approach used by MIT, to detect upper body parts. Position of Driver Vs Standard Front facing position. Plot of Time Vs Position of Neck. Body Pose Estimation for Pedestrians. This allows detecting the dynamics of ‘non-verbal’ communication that happen when a pedestrian crosses the road and looks at the vehicle. Interesting Discovery: Most people look away from the approaching vehicle before crossing the road. Glance Classification: Determining where the Driver is looking. Note: This isn’t the same as gaze detection, where we try to find (x,y,z) pose. We classify two regions: On-Road/Off-Road.Or Six Regions: - On Road- Off Road- Left- Right- Instrument Panel - Rear-View MirrorThe classification allows it to be addressed as a ML problem. The same can be extended to Pedestrian to determine if they are looking at/away from the approaching vehicle. Note: The ground data is provided by manual annotation. Face Alignment: Designing algorithm that are able to detect the individual landmarks of the face and estimating pose of the head. Gaze Classification Pipeline: Source Video Calibration:Determining where the sensor is, since its region based. Video Stabilisation. Face Detection. Face Alignment. Eye/Pupil Detection. Head (and eye) pose estimation. Classification. Decision Pruning. Annotation Tooling:Semi-automated: Data that the network is not confident about are manually annotated. Fundamental Tradeoff:What is the accuracy we are willing to put up with? For increase in accuracy, a human manually iterates and annotates thet data. False Positives:Can be dealt with by more training data. Some degree of human annotation fixes some of the problems. Driver State Detection: Emotion Detection of the driver. Many ways to taxonomize emotion. General Case of detecting emotion. Ex: Affectiva SDK. These algorithms map our expressions to Emotion. Application Specific Emotion Recoginiton:- Ex: Using Voice based GPS Interaction- Self Annotated. - The generic emotion detectors fail here because while driving, ‘smile = frustration’. - Thus Annotation matters. The data must be labelled to reflect these situations. Cognitive Load: Degree to which a person is mentally busy. Eyes expand and contract based on Cognitive load, movements reflect deep thought too. Cognitive load can be detected with blink dynamics, eye movement and pupil dilation. However, in real world lighting leaves out pupil dilation. Blink dynamincs, eye movement are utilised. 3D convolutional NN:- A sequence of images is inputed, we use 3D convolutions.- Convolve across multiple images/channels. - Allow dynamics through time. learning Real World Data:N-back tasks to estimate cognitive load. We detect face, extract eyes and feed these into a CNN. Plot of Eye Movement Vs Cognitive Load. Standard 3D CNN Architecture. Accuracy on real world data. Human Centred Vision for Autonomous Vehicles: Even though we are researching on perception, utilising sensors for localisation and path planning. We are still distant from solving this (Argument: 20+ years). So, Human has to be involved. Thus, the ‘Robot’ needs to understand the ‘Human’s activity’ and the Human-Robot interaction needs to be refined. Path to Mass Scale Automation. (No more steering wheels) Human Centred Autonomy:- A SDC is a personal robot rather than a Perception Control System.- The ‘Transfer of Control’ involves a ‘Personal’ Connection with the machine. - SDCs will be wide reaching. Teaser: MIT SDC will debut on the public streets (Public Testing) in March, 2018. What Next?- DeepTraffic- DeepCrash- SegFuse , connect with me on and are two articles on my Learning Path to You can find me on Twitter @bhutanisanyam1 Linkedin here Here Here Self Driving Cars Subscribe to my Newsletter for a Weekly curated list of Deep learning, Computer Vision Articles

MIT 6.S094: Deep Learning for Self-Driving Cars 2018 Lecture 5 Notes: Deep Learning for Human…

, connect with me on and are two articles on my Learning Path to Self Driving Cars You can find me on Twitter @bhutanisanyam1 Linkedin here Here Here If you want to read more Tutorials/Notes, please check this post out You can find the Markdown File Here These are the notes for the Course (2018), Taught by Lecture 1 MIT 6.S094: Deep Learning for Self-Driving Cars Lex Fridman Lecture 2 Notes can be found here Lecture 3 Notes can be found here Lecture 4 Notes can be found here Lecture 5 Notes can be found here All images are from the Lecture slides. Deep learning: Set of techniques that have worked well for AI techniques in recent years due to advancement in research and GPU capabilities. SDC are Systems that can utilize these. Instructors are working on cars that understand environment inside and outside the car. devoloping Competitions: DeepTraffic: Deep Reinfocement competition, the code runs in browser. v2.0 now allows Multi-Agent training. Learning SegFuse: Dynamic Driving Scene Segmentation competition. Given the Raw video, the kinematics (movements) of automobiles in the video. Training set gives us ground truth labels, pixel level labels, scene segmentation and Optical flow. Goal: To perform better than the State of the art in Image Based Segmentation. Need: Robots need to interpret, understand and track details of scene. Deep Crash: Goal: Use Deep RL to avoid High Speed Crash Avoidance. Training: 1000 runs to train a car model running at 30mph+ speeds with input from a Monocular camera. DeepTesla: Using large scale networks to train end to end steering, using a monocular video as input for training. Why Self Driving Cars? Goal: Applying Data Driven learning methods to autonomous vehicle. It’s a biggest integration of personal robots. Wide Reaching: Huge number of vehicles on road. Profound: An intimate connection between the car and human. Trusting the Robot with your life, a ‘transfer of control’ to the car. The life critical nature of the system if profound and will truly test the limits of the system. Autonomous vehicle: It’s a Personal robot rather than a Perception-Control. The systems will need assistance from humans via a Transfer of control during situations. A truly Perceptional System having a dynamic nature with human equivalence maybe a few decades away. Cognitive load: A fully connected CNN takes in the raw 3D input to analyse the cognitive load,body pose estimation, drowsiness of the driver. Argument: Full Autonomy requires achieving human level Intelligence for some fields. Human-Centred Artifical Intelligence Approach Proposal: Consider human presence in design of every algorithm. Perception control can handle 90% of the cases. Human Control: Dominant in 10% of the cases. Why Deep Learning? Deep learning Perform really well with huge data. Since human lives are directly dependent on the machines, techniques that learn from real world data are needed. Perception/Control Side Human based collobaration and interaction. What is Deep Learning? AI: Ability to accomplish complex goals. Understanding/Reasoning: Ability to turn complex information into simple and useful information. Deep learning (Representation Learning or Feature Learning) is able to take raw information without any meaning and is able to construct hierarchical representation to allow insights to be generated. The most capable branch of AI that is capable of deriving structure from the data to allow to derivation of insights. Representation Learning Representation is important. Ex: Earth Centered Vs Sun Centred. Cartesian Vs Polar Cordinates to seperate circles and triangles Ex: To Seperate Blue Vs Red Curve using a 1 layer hidden Neural Net. The learning (Using raw input to generate outputs) of the function is achieve using Deep Learning Deep learning improves with more data. Generalisation over Edge cases is the main challenge for Deep Learning. Neural Networks. Inspired loosely by biological human neurons. Human NN: 100 billion Neurons, 1,000 Trillion Synapses State of the Art ResNet-52: 60 million Synapses. The difference is a 7 order of magnitude difference. Differences: Human NN have no stacking, ANN are stacked. No Order Vs Ordered Synchronous Learning Vs Asynchronous Learning Unknown Learning Vs Backprop Slower Processing Vs Faster Processing Less Power consumption Vs Inefficient Similiarity: Both are distributed computation on a large scale. A basic Neuron is simple, connected units allow much complex use cases. Neuron: Neuron consists of inputs of a set of edges with weights The weights are multiplied A bias is added A non-linear function determines if the NN is activated. Combination of NN: Feed Forward NN: Successful in Computer Vision. Recurrent NN: Feed back into itself, have memory. Successful in Time Series related data, much similar to Human (hence are harder to train). Universality: Mutiple Neural nets can learn to approximate any function with just 1 hidden network layer* *Given good algorithms. Limitation: Not in power of the Networks, but in the methods. Categories of DL Supervised Learning: Human annotation of data is needed. Augmented Supervised Learning: Human + Machine approach. Semi-Supervised Unsupervised Learning: Machine Input. Reinforcement Learning: Machine Input. Currently being used: 1,2 Future and better categories: 3,4,5. DL Impact Spaces: Defining and Solving a Particular Problem. ex: Boston Housing Price Estimation. General Purpose Intelligence (Or Almost): Reinforcement and Unsupervised Learning. Supervised Learning Training phases: 1. Input Data 2. Labels 3. Training of Data Testing Stage: 1. New data 2. Inputted to learning system 3. Produce Output Learning Forward pass: Input data is fed into the NN and prediction is generated. Backpropagated: Measures deviation from the expected output and calculates error, adjusts the parameters (hyperparameters) used to predict the values according to the magnitude of error. What can we do DL? One to One Mapping. One to Many Many to Many. Asynchronous Many to Many. Terminologies: DL = NN (Deep Learning = Neural Nets). DL is a subset of ML (Machine Learning). MLP: Multi layer Perceptron. DNN: Deep Neural Networks. RNN: Recurrent Neural Networks. LSTM: Long Short Term Memory. CNN: Convolutinal Neural Networks. DBN: Deep Belief Networks. Neural Network Operations: Convolution Pooling Activation Function Backpropogation Activation Functions Sigmoid. Cons: Vanishing Gradient, Not Zero Centred Tanh. Cons: Vanishing Gradients. ReLu. Cons: Not Zero Centred Vanishing Gradients: When the output or gradient of the NN is very low and results in slow learning. Backpropagation Learning Process of the NN. Goal: Update the weights and biases to decrease loss function. Subtasks: Forward pass to compute network output and error. Backward pass to compute gradients. A fraction of the weight’s gradient is subtracted from the weight. Since the process is modular, it’s parallelizable. Learning Learning is an optimization process. Goal: To minimize the Loss Function by updating weights and biases. Techniques used: Mini-batch Gradient Descent and Stochastic Gradient Descent. Learning Challenge Loss function is highly non linear. Vanishing Gradient. Dying ReLU: Derivate = 0 for 0 inputs. Saddle Points. Overfitting: NN Learns training data but fails to generalise well to real world data. Detected by: Low Training error but high Testing Error. Regularization: Techniques that help in generalising. Create Validation set: Subset of Training Data Early Stoppage: To save a checkpoint and evaluate how is the NN performing on Test Data. Dropout: Randomly Remove some of the nodes (along with the incoming and outgoing nodes) Denoted by Probability of keeping a node (p) Input Node p should be much higher. Goal: To help generalise better. Regularization Weight Penalty: L2 Penalty: Penalised squared weights: Keeps smaller weights unless error derivative is high. Prevents from fitting sampling error. Smoother Model. For 2 similar inputs, the weight gets distributed. L1 Penalty: Penalise absolute weight: Allows weights to remain large. Neural Network Playground: To play around with techniques and practise Deep Learning Breakthroughs What Changed? Compute Power increased. Large Organised Datasets available. Algorithms and Research in GPU Utilization. Software and Infrastructure. Financial Backing. DL is hard Human Comparision: Human Vision: Devoloped 540,000,000 years of data. Bipedal Movement: 230,000,000 years of data. Abstract Thoughts: 100,000 years of data. Neural Net: Add Distortion to pixel data, causes incorrect predictions. Vision Problems: Illuminabilty, Poses, Occlusions, Intra class variations. Object Recognition/Classifcation: Goal: Input an image and predict an output ImageNet: 14M+ categories with 21.8k+ Categories Competition: ILSVRC: AlexNet (2012) made a significant jump in accuracy. Resnet (2015): Human Level Performance was defeated. Subtle example: DL is still distant from ‘Human Generalisation Ability’ Same Architecture, Many Applications: We can change the predict layer to make predictions to as many number of classes as per requirements. Image Classification. Image Captioning. Object Localisation. Image Segmentation. FCNN: Every pixel is assigned a class and it inputs an image and produces another image as output. Goal: Image to Image Mapping. Use-Cases: Pixel-Level Full Scene Segmentation. Colorisation Mapping. Object Detection. Background Removal. Pix2PixHD: Generate High Res Photo Realistic images from semantic label maps. RNN: Work with Sequences Use-Cases: Handwriting Generation Image Caption Generation. Video Description Generation. Modeling Attention Steering. Drawing with Selective Attention. Major Break Throughs Pong to Pong (2012): A step towards AGI. AlphaGo (2016): Learned from Human Expert Games AlphaGo Zero (2017): Beat AlphaGo and Co and it was trained without any Data! (It learned by practising against itself). DeepStack (2017): First to beat Professional Poker Players (in Heads up Poker) Current Drawbacks Defining good Reward Function is difficult. (Coast Runner Examples), The outcomes can be surprising. Lack of Robustness: Adding noise to Pixels cause wrong predictions. Current Challenges: Transfer Learning: Works well with domains closely related. Challenge: Cross-Domain Transfer learning is missing. Reason: Understanding of reasonsing or ability to derive understanding Huge Amount of data needed. Annotated Data is required. Not fully automated: Hyperparameter tuning Reward: Defining a good reward function is difficult. Transparency: NN are mostly blackboxes (even after we visualise the under the hood processes). Edge cases: DL is not good at dealing with edge cases. (Specially when it comes to Autonomous driving). Re: Why DL? It’s an oppurtunity to apply techniques to real world problems effectively. (And DL is the most effective at these). , connect with me on and are two articles on my Learning Path to Self Driving Cars You can find me on Twitter @bhutanisanyam1 Linkedin here Here Here Subscribe to my Newsletter for a Weekly curated list of Deep learning, Computer Vision Articles

Stories, News & Mentions about DEEPTRAFFIC

Deeptraffic’s Stories on HackerNoon