<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by John Anthony Radosta on Medium]]></title>
        <description><![CDATA[Stories by John Anthony Radosta on Medium]]></description>
        <link>https://medium.com/@jradosta?source=rss-c21bcd7f76c0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*F6-dlwg2qNXjNbqPSZ0Qlg.jpeg</url>
            <title>Stories by John Anthony Radosta on Medium</title>
            <link>https://medium.com/@jradosta?source=rss-c21bcd7f76c0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:37:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jradosta/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Quantum Nature of Truth]]></title>
            <link>https://jradosta.medium.com/the-quantum-nature-of-truth-703c1d435911?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/703c1d435911</guid>
            <category><![CDATA[philosophy-of-life]]></category>
            <category><![CDATA[quantum-mechanics]]></category>
            <category><![CDATA[quantum]]></category>
            <category><![CDATA[philosophy-of-mind]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Mon, 10 Feb 2025 20:30:17 GMT</pubDate>
            <atom:updated>2025-02-10T20:34:33.803Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Things are not what they seem…but they also are.</em></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_J8lc8P4-gHdWmIYF5gYkA.jpeg" /></figure><p>I’m reading a book about behavioral profiling and the author mentioned how eye blink rate is in an indicator for lawyers profiling prospective jurors to understand which jurors they may be able to persuade and which jurors they will need to ‘work on’.</p><p>This last sentence sent my mind on a whole tangent about the nature of reality and what a criminal trial indicates about the nature of reality.</p><p><strong><em>Because distilled down—what is a criminal trial after all?</em></strong></p><p>A criminal trial is a proceeding in which a group of people with opposing interests (prosecution and defense) fight to establish their version of the truth into a form that they believe 12 random people will agree upon.</p><p><strong><em>It’s literally a game of adversarial beliefs to determine ‘the truth’ where the first to 12 wins.</em></strong></p><p>And therein lies the interesting part and the main idea behind my tangent:</p><blockquote>Isn’t it crazy that the truth is so subjective and elusive that a court of law requires 12 people to agree on it?</blockquote><p>Not one…not two…not four…not even ten…<em>but 12 people!</em></p><p>And this lead me down another rabbit hole thought about the nature of reality which is that if <em>12 people</em> are required in a legal setting to decide on the truth about something, is there really such a thing as <em>objective</em> truth in the classical sense with anything?</p><p>The obvious answer to that question is ‘No.’</p><p><strong><em>So if there’s no such thing as an objective truth…then what are 12 jurors actually agreeing upon in a court of law?</em></strong></p><p>In order to answer this, I realized that we cannot look at truth from a classical standpoint anymore, we have to look at it from a quantum standpoint.</p><h3><strong>The Quantum Nature of Reality</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*meRzGlZ5kE50f8v-GIsaLw.jpeg" /></figure><p>You cannot look at truth without considering probabilities, because everything is something until it’s not and the only constant in the universe is change.</p><ul><li>Relationships are great, until they are not.</li><li>It is daytime outside, until it becomes night time.</li><li>The economy is booming, until it starts to falter and vice versa.</li></ul><p><strong>The state of everything is a quantum pendulum that is in constant motion and unknown until the exact point of measurement.</strong></p><ul><li>The economy is officially in a recession after two declining quarters of real GDP.</li><li>It become nighttime tonight at 6:10PM EST.</li><li>Your relationship was perfect, up until someone cheated on the other person.</li></ul><p><strong><em>The point I am making is that every ‘fact’ is a pendulum in motion, and at any given moment, the value of that pendulum is an unknown value until it is directly and consciously observed.</em></strong></p><p>This is the nature of reality at the quantum level as well, where a particle’s position is not truly known until it directly scrutinized.</p><p><strong><em>Which is what I’m starting to realize about all truths…</em></strong></p><p>Truth is not a constant ‘thing’ or fact, it is something that upon observation and scrutinization, begins to collapse into an observable and measurable belief much like a quantum wave function collapses into a measurable particle position.</p><p>A truth is based on facts, but facts are debatable things because they are all pendulums in motion.</p><p>A fact by itself in a vacuum doesn’t tell us much. It is only when stable facts chained together in a logical chain reveal a stable and measurable form of reality that we call ‘truth.’</p><p>This is much like how transistors in a classical computer work as well, a transistor by itself can only provide so much logic:</p><ul><li>0 or 1.</li><li>Yes or No.</li><li>True or False.</li></ul><p>But transistors taken in groups of 0’s or 1’s can encode more logic.</p><ul><li>01000001 is the letter “A”</li><li>00111111 is a question mark.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H2m24gGhxu3klfA7NSbqdg.png" /></figure><p><strong><em>The truth about anything works pretty much the same way. It is only when we string together multiple observed facts into a chain of ‘factual binary’ that a truth can be ‘encoded’ onto reality.</em></strong></p><p>Let’s observe this phenomenon from the perspective of a criminal court case.</p><p>Let’s pretend we have a murder trial in New York and we have the following evidence being presented to the court.</p><ul><li>The defendant was videotaped by a security camera entering and leaving the scene of the crime between 7AM and 9AM.</li><li>The coroner’s report showed the time of death of the victim to be between 7AM and 9AM on the same date.</li><li>A receipt for the purchase of the murder weapon by the defendant was presented.</li><li>DNA of the defendant was found on the victim’s body at the scene.</li><li>The defendant and the victim had an argument in which the defendant discovered that the victim was in possession of information that would put the defendant into serious financial difficulties.</li></ul><p><strong>So here we have:</strong></p><ul><li>Evidence of the suspect at the scene of the crime during the time it occurred.</li><li>The DNA of the defendant on the victim.</li><li>A direct link to the murder weapon.</li><li>A clear and distinct motive.</li></ul><p><strong>Based on these facts, would you convict this suspect?</strong></p><h3>Shifting the ‘QuBits’ of a Murder Trial</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*CM7gc4C3RyD32qMIIKFWYg.jpeg" /></figure><p>Now using the same example, let’s read the same ‘QuBits’ of information above in way that will show us the quantum nature of even something as seemingly definitive and objective like a murder.</p><ul><li>The defense presents information that the clock on the security camera was actually in UTC time, while the coroner’s report on the time of death was in Eastern Standard Time, a 5-hour timezone difference.</li><li>This means that the security camera’s footage timestamp of the defendant entering the scene of the crime at 7AM and 9AM in UTC was really at 2AM-4AM EST.</li><li>The suspect and the victim were married.</li></ul><p>So using the same QuBits of information, we have a vastly different story of truth:</p><ul><li>The defendant and victim being lovers explains the potential for DNA evidence already being on the victim at the scene of the crime.</li><li>The camera footage time conversion shows the defendant had already left the scene of the crime 5 hours before the coroner’s estimated time of death.</li><li>The defendant and the victim being married weakens the motive as the defendant and victim’s financial scenarios were aligned together, and couples fighting over finances is common.</li></ul><p>So here we have basically the same QuBits of information from above, but their observed values have changed.</p><p><strong>Do you think the defendant is guilty now?</strong></p><p>Better question…does this mean the defendant is innocent?</p><p>Not necessarily!</p><p>All we have now is a vector of ‘truth’ that gives plausible deniability for a conviction by 12 jurors.</p><h3>All Truths Are Quantum Truths</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DCjlZiSmq7IQ7-uqfNhGAA.png" /></figure><p>The murder trial example above and the nature of the 12-man jury in the legal system show us that the ‘truth’ about anything is dependent upon the vector of facts you directly observe to get there—some vectors are inherently stronger or more efficient than others.</p><p><strong>This has a very profound implication and conclusion:</strong></p><blockquote>You can literally ‘think’ anything into existence if your thought vector is strong enough.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=703c1d435911" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Great Tech Products Don’t Solve Problems—They Do So Much More]]></title>
            <link>https://jradosta.medium.com/great-tech-products-dont-solve-problems-they-do-so-much-more-059f2b99652b?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/059f2b99652b</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Thu, 15 Aug 2024 17:24:25 GMT</pubDate>
            <atom:updated>2024-08-15T18:20:51.826Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/843/1*2j1yk4MPAzwndHBjKxhf2w.jpeg" /></figure><p>Product teams in the tech industry are focused on shipping products with various bells and whistles without any concern as to whether it actually does anything measurable for the customer’s business.</p><p>I can certainly attest to this as an AI engineer, everybody wants a chatbot all of sudden.</p><p>But why? How does a chatbot add revenue or reduce costs?</p><p>I would even go as far as to say it could hurt both.</p><p>Automated phone bots have already frustrated end users for years — every person reading this has had the experience of repeatedly pressing “0” or saying “representative” until we can talk to a live person.</p><p>I’m at the point of my career where I’ve realized that great technology does not solve problems. It never has.</p><p>It has to do so much more.</p><h3>Technology Doesn’t Solve—It Enables</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sTb2GYwvRnW4JLqQjC92RA.jpeg" /></figure><p>Look at the Wright Brothers and the first flight.</p><p>Did their plane solve the “problem” of humans not being able to fly?</p><p><strong>Nobody prior to 1903 ever had the following conversation:</strong></p><blockquote>“Hey Joe, what seems to be bothering you?”</blockquote><blockquote>“(Looking up at the sky) Well, Bill…I’m trying to get to New York, and I just can’t seem to flap my arms hard enough to get there!”</blockquote><p>It’s a ridiculous premise, but that is what we get when we look at technology as a solving problems.</p><p><strong>Technology has never solved a problem. Ever.</strong></p><p>Technology enables us to do things that never even occurred to us as possible.</p><p><strong>It gives us a new ability—it doesn’t fix an old one.</strong></p><p>That is why creating and shipping technology products is so hard — you have to reimagine the human experience of something entirely…</p><p>To borrow from Disney, you have to become an <em>imagineer</em>.</p><blockquote>Great technology gives us a new ability—it doesn’t fix an old one.</blockquote><p>Every time a good technology product has come along, it shifted a paradigm and changed the way we did something entirely.</p><ul><li>The airplane changed the way we travel — it didn’t solve the problem of humans not having feathers and wings.</li><li>Social media platforms changed the way we maintain relationships — it didn’t solve a problem with the Rolodex.</li><li>ChatGPT changed the way people search for answers — it did not solve the problem of search engines not being conversational.</li></ul><p>Technology allows us to perceive the world in a different way. It changes our paradigm and how we think about the world.</p><p>Let’s look at an example of how great technologies change how we think.</p><p><strong>Q: Let’s go back to 1900 — if I asked you to calculate the time it would take for you to arrive in New York City from some random city in America, what would you have done?</strong></p><p>A: You probably would’ve pulled out a map, figured out what trains you’d need to catch and the scheduled arrival date of each train to come up with an answer.</p><p><strong>Q:Fast forward to today — if I asked you to do the same thing, what would you do?</strong></p><p>A: You’d go online and check flights from your nearest airport and maybe factor in the time for the Uber rides as well.</p><p><em>You wouldn’t calculate the time to get to New York City as the crow flies.</em></p><p>That is because your paradigm for travel is limited by the technology you have available to you—you perceive the world relative to your limitations of perceiving it.</p><p>You currently do not have the paradigm of personal air travel enabled to you, such as a jet pack or a flying car.</p><p>Therefore, your paradigm for travel planning comes down to airports and Ubers.</p><blockquote><em>You perceive the world relative to your limitations of perceiving it.</em></blockquote><p>When you look at technology as an enabler and mechanism for shifting paradigms instead of solving problems, you begin to see it for what it truly is.</p><p><strong>Great technology products give you a new reality—they don’t band-aid your current one.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=059f2b99652b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[This Investing Strategy Works 90% Of The Time—Here’s How To Use It]]></title>
            <link>https://jradosta.medium.com/this-investing-strategy-works-90-of-the-time-heres-how-to-use-it-f9420a2044f2?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/f9420a2044f2</guid>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[money]]></category>
            <category><![CDATA[money-management]]></category>
            <category><![CDATA[investing-tips]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Mon, 15 May 2023 14:18:34 GMT</pubDate>
            <atom:updated>2023-05-15T14:47:36.324Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/916/1*EvDaVfW4xf_A6K4N6d3Sag.jpeg" /></figure><p>The <strong>Benner Cycle</strong>, created by Samuel Benner, has consistently proven to work over time, with a success rate of 90%.</p><p>The Benner Cycle is based on the idea that the stock market moves in predictable patterns that can be forecasted…<strong>[see the full strategy below]</strong></p><p>This Investing Strategy Has a 90% Successful Track Record — Here’s How You Can Use It…</p><p>The Benner Cycle is based on the idea that the stock market moves in predictable patterns that can be forecasted…[read more below]</p><p><a href="https://learn.synvestable.com/benner-cycle/">This Investing Strategy Has a 90% Successful Track Record - Here&#39;s How You Can Use It</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f9420a2044f2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Biggest Hurdle Facing Decentralized Finance (DeFi)]]></title>
            <link>https://medium.com/coinmonks/the-biggest-hurdle-facing-decentralized-finance-defi-1e1151c4afe?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/1e1151c4afe</guid>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[bitcoin]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[defi]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Mon, 01 Mar 2021 17:56:22 GMT</pubDate>
            <atom:updated>2024-10-23T18:38:22.950Z</atom:updated>
            <content:encoded><![CDATA[<p>Last month, I provided a solution for <a href="https://chain.link/">Chainlink</a> on how best to model implied volatility (IV) on-chain for both Bitcoin and Ethereum options using price feeds and oracles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZmvilvAkFgO5TKJHfsjLtw.png" /><figcaption>Chainlink — Decentralized Pricing Feeds For Etherum Smart Contracts</figcaption></figure><p>If you’re not super familiar with how Ethereum smart contracts work or <a href="https://blog.chain.link/what-is-the-blockchain-oracle-problem/">The Oracle Problem</a>, I’ll give a brief summary <em>before</em> diving into one of the major issues facing public blockchain protocols.</p><h3><strong>The Isolation Problem of Public Blockchain</strong></h3><p>Public blockchains are gaining popularity because of the their emphasis on both decentralized infrastructure and payment authorization. The premise is that decentralization provides fairness when it comes to data sharing, payments, and other industry applications which are typically offered as a service by a central authority — like Visa, Mastercard, Google, or Paypal.</p><p>While blockchains certainly offer a unique solution to the decentralization problem, they have unique issues surrounding off-chain data ingestion that can introduce security risks, fraudulent transaction, and just plain wrong application state that becomes immutable (non-removable) within the history of the blockchain ledger.</p><p><strong>In other words, blockchains have a data problem.</strong></p><p>So the question becomes:</p><blockquote>How do you consume data from the real world, and how do you know it can be trusted so that transactions added to the ledger are both accurate and fair?</blockquote><h3>Data Problems Become Oracle Problems</h3><p>To sum it up, the data problem with blockchains revolves around a very simple limitation — <strong>blockchains cannot pull in data from or push data out to any external system as a built-in functionality.</strong> <strong>As such, blockchains are isolated networks very akin to a computer with no Internet connection.</strong></p><p>Companies like <a href="http://chain.link/">Chainlink</a> are attempting to solve this problem by offering prebuilt solutions known as <em>price feeds</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PVocV01ntSKIk0BIbA70TA.png" /><figcaption>Price Feed Example for ETH / USD Exchange Rate Using Various Sources,</figcaption></figure><p>These price feeds are essentially oracles that collect data off-chain and aggregate them into a single-source of truth using a weighted average.</p><p>While this works well with price feeds and raw data, the major issue facing smart contracts currently is that higher order financial modeling becomes nearly impossible. And if blockchains’ ultimate goal is to decentralize finance (DeFi), then it’s a problem that will need to be solved going forward.</p><h3>The Disconnect Between Traditional Finance &amp; DeFi</h3><p>The issue that I see facing public blockchain protocols when it comes to DeFi isn’t necessary a philosophical problem (centralization vs decentralization), it’s a technical problem.</p><p>When it comes to risk management or portfolio allocation in traditional finance (TradFi), fund managers often use financial models to make portfolio allocations that quantify risk in a variety of manners. These risk management metrics often take the form of greek letters: <em>alpha, beta, delta, gamma, rho, theta, etc. </em>which allow managers to measure portfolio sensitivities to market changes in underlying prices, market volatility, interest rates, and other factors.</p><p>In addition to this, quantitative money managers often use bayesian modeling to model portfolio performance as a distribution of probabilities, so these greek letters may take on values of multiple probability distributions summed together (Bayesian statistics).</p><blockquote>The major issue facing blockchains right now is that if you’re going to truly decentralize finance, you have to make available the same sets of tools that are available to traditional finance, otherwise all you’re doing is taking on risk that you can’t quantify.</blockquote><h3>Current Limitation of Smart Contracts</h3><p>In the previous section, I mentioned that risk management and thus portfolio management often boils down to greek letters. These greek letters are usually partial derivatives of a variety of financial modeling equations, the most famous one being the Black-Scholes Model for option pricing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*94nQ0xHn-Ikv-rT_3NsOqw.png" /><figcaption>The closed-form Black-Scholes Equation for option pricing. The equation is used for deriving the expected price of an option based on the price of the current price of underlying asset, the cost of capital, the time to expiration, the strike price, and the asset’s volatility.</figcaption></figure><p>Here, we see the problem facing blockchains when it comes to TradFi modeling, which is the fact that oftentimes, TradFi models include distributions. In the case of the Black Scholes Model, that “N” symbol is a log-normal distribution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7mWi10xb5VdQwuo8lrCUdg.jpeg" /><figcaption>The log-normal distribution, where Y is function of X distributed over a range of probabilities.</figcaption></figure><p>Which brings us to our most crucial and technical problem facing smart contracts today…</p><p><strong>The major problem with Ethereum smart contracts is that Solidity as a programming language does not support floating point [decimal] numbers</strong>, which makes manual calculations of implied volatility with something like the Black Scholes Merton (BSM) Model within a smart contract impossible.</p><p>If you don’t have floating point numbers, you can’t model continuous distributions, which means you can’t do TradFi modeling on-chain. This is a major hurdle facing Solidity in terms of adoption for decentralized asset management.</p><p>Comparatively, all modern programming languages such as Python, Java, C++, Javascript, Go, and even Rust all support floating point numbers.</p><p>Python is right now <em>the defacto</em> language for financial modeling because of it’s extensive availability of finance libraries such as TA-Lib and Mibian. If independent financial professionals and developers for independent quantitative asset managers are going to adopt smart contracts as a viable transparent money management solution, then this floating point number problem will need to be solved.</p><p>And these libraries will need to be rebuilt on-chain because nobody wants to waste time doing Black-Scholes calculations by hand. Developers want to focus on solving real world problems, not reinventing the wheel.</p><h3>The Only Viable Current Solution</h3><p>The only solution that’s available and scalable in terms of performance right now for doing TradFi modeling with Ethereum is to do all that modeling <em>off-chain</em> in a midlayer API, round the result to an integer or an array of integers, and return that result to the calling oracle.</p><p>Besides the inability to do this modeling on-chain because of the lack of floating number support in Solidity, even if you tried, the gas prices would probably kill your transaction because of the number of calculations needed under the hood — there’s a lot of <em>sigmas</em> involved with calculating distributions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/225/1*qXKII4FQIQLEqhg9eT0IMw.png" /><figcaption>Calculations involving sigmas could cause gas prices on Ethereum to explode.</figcaption></figure><p>In the Colab Notebook below, I provided a working example of how TradFi modeling would need to be done in Ethereum’s current version of Solidity.</p><p>As an example for Chainlink, I showed how you would model implied volatility (IV) for Bitcoin and Ethereum option contracts using Chainlink price feeds.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*miCo2Ypt1NC4Y4gsC-fmrQ.png" /></figure><p>In short, you need to do all the heavy lifting in the midlayer API, and then cache or stash the result as rounded integers on-chain, in which case the smart contract more or less is just serving as a database.</p><blockquote>But this introduces a whole plethora of philosophical issues that run counter to blockchain’s premise: if the financial modeling itself isn’t decentralized, what good is decentralizing the results of those models?</blockquote><p>This is a major issue that the Ethereum Foundation will need to address because this is the direction in which all of finance is moving: <em>probabilistic risk management.</em></p><p>It’s also the intersection of where probabilistic deep learning and finance are beginning to work together, so deep learning frameworks like TensorFlow will also need to come on-chain as well.</p><p>If smart contracts cannot eventually offer this basic floating point number functionality to do TradFi modeling on-chain, how can you truly<em> </em>expect to decentralize <em>all</em> of Western finance?</p><h3>Practically Decentralizing Western Finance</h3><p>As I write finish writing this article, I’m looking at my Starbucks coffee to the right of my laptop. It cost me $1.86 today, a decimal point number.</p><p>If Starbucks’ POS system was built in Solidity, they’d have to round that number and charge me $2 exactly. Finance doesn’t work that way in the real world.</p><p>We still have a ways to go before we decentralize and democratize the financial world completely. But I trust the Ethereum Foundation developers are up to the task of bridging this gap between TradFi and DeFi.</p><h3>If you learned something from this post, give me some claps and a follow…</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/694/1*lMmW_WjZwZzKjOGuU0975g.gif" /></figure><h3>And if you need something done right…</h3><p><a href="https://www2.synvestable.com/">Synvestable — Your Global S&amp;P 500 Consulting Partner</a></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FF8bCn3mBZVc%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DF8bCn3mBZVc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FF8bCn3mBZVc%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/941427d93a81ed5c2696cc0b4685f042/href">https://medium.com/media/941427d93a81ed5c2696cc0b4685f042/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1e1151c4afe" width="1" height="1" alt=""><hr><p><a href="https://medium.com/coinmonks/the-biggest-hurdle-facing-decentralized-finance-defi-1e1151c4afe">The Biggest Hurdle Facing Decentralized Finance (DeFi)</a> was originally published in <a href="https://medium.com/coinmonks">Coinmonks</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Google’s Professional AI Certification & What I’ve Learned Since]]></title>
            <link>https://becominghuman.ai/googles-professional-ai-certification-what-i-ve-learned-since-40489b96c6d9?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/40489b96c6d9</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[tensorflow]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Mon, 22 Feb 2021 14:09:45 GMT</pubDate>
            <atom:updated>2021-03-01T14:48:47.603Z</atom:updated>
            <content:encoded><![CDATA[<p>Last week, I took Google’s Tensorflow Certification exam, a grueling 5-hour endeavor where the tester is required to build highly accurate models pertaining to image classification, text classification, and time-series predictors.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fMK7Y6XKoeHlJ5o9wGrMVQ.png" /></figure><p>I ended up passing the exam, but it definitely was a beast and required incremental accumulation of subject matter knowledge over several years along with a very structured study schedule and hands-on practice.</p><p>I wanted to quickly share what I’ve learned since then about Tensorflow and about neural networks in general.</p><h3>Artificial Intelligence Is Probabilities</h3><p><a href="https://becominghuman.ai/ai-machine-learning-deep-learning-explained-in-5-minutes-b88b6ee65846">Artificial Intelligence</a> (AI) is all about probabilities at the most basic level, but we can sum it up in once sentence:</p><blockquote><strong>Artificial Intelligence is about arriving at the probability distribution of the truth.</strong></blockquote><p>You aren’t seeing this at the surface level because API’s like Keras extract the higher level math away, but ultimately its good to understand exactly what machine learning is doing under the hood so that we know when to put faith into what the model is telling us, and when not to.</p><p>So how are neural networks using probability distributions? Let’s take an example of using an <a href="https://becominghuman.ai/google-will-beat-apple-at-its-own-game-with-superior-ai-534ab3ada949">AI </a>model to predict heart disease.</p><figure><a href="https://aijobsboard.com/?fbclid=IwAR0GcsLs8A6CFM-fynrZMM3sTgo_Zpbto2CjglJm0Dmi6otC6YWc7CDFpQk"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dAlwXfVrtYspdCzmC8iADw.png" /></a><figcaption>Big Data Jobs</figcaption></figure><p>Let’s say I have a training dataset of 1000 patients, 980 are labeled to not have heart disease and 20 are labeled as having heart disease. As far as the model is concerned, this dataset represents the <em>ground truth, </em>the maximum amount of information <strong>present</strong> to be learned by a machine learning model to equate the real world.</p><p>What a neural network is doing internally is that it’s shifting its weights around to arrive at the expected distribution of the original training dataset. In other words, when you feed data into a neural network, what you’re saying to the network is this:</p><blockquote><em>“Okay, neural network, I’m going to feed you some data where 980 of the inputs have class “0&quot; (“do not have heart disease”), and 20 have class “1&quot; (“have heart disease”). Please shift your network weights around when evaluating the input data so that your predictions will typically match around 20 class “1’s” for every 980 class “0’s””.</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I-MxKoiWxJXLfExpvbT1eQ.png" /><figcaption>“Okay Neural Network, given all these X’s, give me the function for a red line that best fits the data distribution so I can predict future Y’s”</figcaption></figure><p>That’s all that’s happening under the hood. The neural network is converging in on a probability distribution for predictions that will most closely match the probability distribution of the original ground truth you fed into it.</p><p>This is an important thing to understand because if you’re feeding datasets into a machine learning model where the distribution of classifiers do not approximate the <em>true distribution </em>of those classifiers in nature, then your machine learning model will be pretty “untruthful” (useless) when you try to predict on new data.</p><p>This is the ultimate challenge of data science and a<a href="https://becominghuman.ai/6-ai-subscriptions-to-keep-you-informed-4cfc6cb024a">rtificial intelligence </a>is that it’s not just about building efficient models, it’s actually more about what you’re feeding into them.</p><p>This is analogous to diet and exercise for people: <em>if you’re training hard in the gym, but eating fattening and unhealthy foods, it doesn’t matter how great your training program is, you’re ultimately not going to get the results you’re looking for.</em></p><p>The hardest part about machine learning is understanding whether the distribution of your data matches that which occurs naturally in the physical world.</p><h3>Trending AI Articles:</h3><blockquote><a href="https://becominghuman.ai/top-5-open-source-machine-learning-recommender-system-projects-with-resources-50583c4007c3">1. Top 5 Open-Source Machine Learning Recommender System Projects With Resources</a></blockquote><blockquote><a href="https://becominghuman.ai/deep-learning-algorithms-in-self-driving-cars-14b13a895068">2. Deep Learning in Self-Driving Cars</a></blockquote><blockquote><a href="https://becominghuman.ai/generalization-technique-for-ml-models-ed6ba666d171">3. Generalization Technique for ML models</a></blockquote><blockquote><a href="https://becominghuman.ai/why-you-should-ditch-your-in-house-training-data-tools-and-avoid-building-your-own-ef78915ee84f">4. Why You Should Ditch Your In-House Training Data Tools (And Avoid Building Your Own)</a></blockquote><p>If I went out and sampled 1000 people for heart disease using their age, BMI, blood pressure, and other feature inputs to predict whether they have heart disease, how many of them would <em>truly</em> have heart disease?</p><p>But here’s the twist:</p><ul><li>What if I sampled those 1,000 people standing outside my local gym? How might that affect the number of people in the dataset I record as truly having heart disease?</li><li>Conversely, what if I sampled the entire dataset standing outside of a McDonald’s?</li></ul><p>Depending on how I collect the dataset, that is going to affect the ground truth distribution of “doesn’t have heart disease” versus “has heart disease” for training data. I may encounter far less people who truly have heart disease standing outside my local gym versus people who are regular diners at McDonald’s.</p><p>If I fed either of these datasets into a machine learning model, the model is going to converge in on the same probability distribution as the dataset that I fed into it. The model cannot learn any more information that what I originally gave it. It won’t tell me whether I am biasing it or not.</p><blockquote>Neural networks cannot learn any more information than what you give them. They cannot tell you whether your dataset represents actual reality or not.</blockquote><p>If my dataset is biased towards not having heart disease, the model will move its internal probability distribution to put a heavier emphasis on predicting someone to <em>not</em> have heart disease, which could yield a lot of false negatives if I started using that model in a clinical setting for cardiology patients.</p><p>This is fundamentally how artificial intelligence works: <em>arrive at the probability distribution that would yield all these “y’s” for all these “x’s”</em>. It’s an important concept to understand not just from a results perspective in machine learning, but also from an ethical standpoint.</p><p>Ultimately, a<a href="https://becominghuman.ai/write-your-first-ai-project-in-15-minutes-a287ed348e7c">rtificial intelligence</a> will give back to us only what we feed into them, they are merely a mirror for our own cognitive biases. It’s important that we realize that as we progress into the future.</p><h3>Don’t forget to give us your 👏 !</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/780/0*2lvCls4yjxVMfZSR" /></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F8f5f8b%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F8f5f8b&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/c43026df6fee7cdb1aab8aaf916125ea/href">https://medium.com/media/c43026df6fee7cdb1aab8aaf916125ea/href</a></iframe><figure><a href="https://becominghuman.ai/artificial-intelligence-communities-c305f28e674c"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*2f7OqE2AJK1KSrhkmD9ZMw.png" /></a></figure><figure><a href="https://upscri.be/8f5f8b"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*v-PpfkSWHbvlWWamSVHHWg.png" /></a></figure><figure><a href="https://becominghuman.ai/write-for-us-48270209de63"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*Wt2auqISiEAOZxJ-I7brDQ.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40489b96c6d9" width="1" height="1" alt=""><hr><p><a href="https://becominghuman.ai/googles-professional-ai-certification-what-i-ve-learned-since-40489b96c6d9">Google’s Professional AI Certification &amp; What I’ve Learned Since</a> was originally published in <a href="https://becominghuman.ai/">Becoming Human: Artificial Intelligence Magazine</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Are The Financial Markets Better Off Today Than 4 Years Ago?]]></title>
            <link>https://jradosta.medium.com/are-the-financial-markets-better-off-today-than-4-years-ago-310b5dc008fc?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/310b5dc008fc</guid>
            <category><![CDATA[biden]]></category>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[financial-markets]]></category>
            <category><![CDATA[trump]]></category>
            <category><![CDATA[investing]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Thu, 16 Jul 2020 00:47:07 GMT</pubDate>
            <atom:updated>2020-07-16T00:47:07.034Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*1eO6JFE-z9Qg4AZDmJKqhg.jpeg" /><figcaption>The results may surprise you.</figcaption></figure><p>I wanted to quickly share some of my original research from today on the markets since we’re getting into the thick of things going into November.</p><p>Are the financial markets better off now than 4 years ago?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*--npO6QAhWh7hRI-RrZquA.png" /></figure><p>The S&amp;P 500 since 2014 has had essentially the same distribution of expected daily returns (fat-tailed normal) as it has for the last 50 years.</p><p>However, investors (you and I) have had to take on more risk for the same expected distribution of returns as intraday market volatility has significantly increased beginning in 2018.</p><p>So are the financial markets better off than they were 4 years ago?</p><p>Well, we’re about the same in terms of daily upside and downside expectations, only we’ve had to put up with much wilder and volatile swings getting there.</p><p>In other words…s<em>ame returns, but higher risk </em>as the waters have become choppier.</p><blockquote>Connect with me on <a href="http://www.linkedin.com/in/johnradosta">LinkedIn</a>.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=310b5dc008fc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Markets Are Like An Ocean]]></title>
            <link>https://jradosta.medium.com/markets-are-like-an-ocean-1b0eafa87b0c?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b0eafa87b0c</guid>
            <category><![CDATA[market-research]]></category>
            <category><![CDATA[quantitative-strategies]]></category>
            <category><![CDATA[market]]></category>
            <category><![CDATA[marketing-strategies]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Fri, 31 Jan 2020 22:19:34 GMT</pubDate>
            <atom:updated>2020-01-31T22:19:34.680Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*-Sgdb-Gv_VUcjGnqnhiZrQ@2x.jpeg" /></figure><p>I’m really starting to see the markets like a great ocean — there are parts of the ocean that are turbulent, other parts, in the very same moment, are calm.</p><p>You survey the parts of the ocean your curious about based on the logs of past journeys…you plot your course, your routes, and you know what to expect from potential dangers.</p><p>Then once you’ve plotted your route and your potential escape routes, you just embark — and trust your map of the territory.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b0eafa87b0c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[In Honor of Kobe Bryant]]></title>
            <link>https://jradosta.medium.com/in-honor-of-kobe-dac764abaa67?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/dac764abaa67</guid>
            <category><![CDATA[opinion]]></category>
            <category><![CDATA[motivation]]></category>
            <category><![CDATA[kobe-bryant]]></category>
            <category><![CDATA[kobe]]></category>
            <category><![CDATA[news]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Tue, 28 Jan 2020 04:33:30 GMT</pubDate>
            <atom:updated>2020-01-28T04:44:00.678Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/810/1*MW6WGS7XcGb2WFtiSA4FGA.jpeg" /></figure><p>I haven’t stopped thinking about Kobe Bryant today. Been thinking about how such a tragedy could occur, the meaning behind it…and then I realized that’s not the way to think about it.</p><p>When bad things happen, its important to find the silver lining, and the silver lining I see is people all over the world, coming together to celebrate this man’s life and greatness. He truly was a model for excellence and that model will live on for generations to come.</p><p>This is one of my favorite interviews of his and the mindset that made him great.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FG5ObUxVVnhM%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DG5ObUxVVnhM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FG5ObUxVVnhM%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/f8b8ad819d813a273473ab76f5c452cb/href">https://medium.com/media/f8b8ad819d813a273473ab76f5c452cb/href</a></iframe><h3><strong>Key Takeaways:</strong></h3><p>1) Be competitive. You may not be everybody’s friend, but they will respect you and you will bring the best out of them.</p><p>2) Play to your weaknesses. Get better at the areas where you struggle that are important to your game. Do not shirk the responsibility of having to improve yourself.</p><p>3) Get over yourself. Public failure isn’t about your ego. Suck it up and analyze the situation for what it was. Ignore the commentary and get to the root of the problem.</p><p>4) When the game is on the line, play with a sense of urgency. Forget about your own pain, you can worry about that after. Focus on the moment at hand, and thrive through it.</p><p>5) Define your life on your own terms. Circumstances do not define you, you define you. Period.</p><p>#ThankYouKobe</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/487/1*hXC6_ZfbfRO-bqRn_Q-Z8w@2x.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dac764abaa67" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Containers Or Serverless? The Battle For Your DevOps]]></title>
            <link>https://jradosta.medium.com/containers-or-serverless-the-battle-for-your-devops-4288fc999d18?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/4288fc999d18</guid>
            <category><![CDATA[serverless]]></category>
            <category><![CDATA[cloud]]></category>
            <category><![CDATA[containers]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Fri, 15 Nov 2019 17:51:56 GMT</pubDate>
            <atom:updated>2019-11-15T17:52:09.641Z</atom:updated>
            <content:encoded><![CDATA[<h3>Containers Or Serverless? The Battle For Your DevOps Mindshare</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*syXCcRxyUek2-705yRXueg.jpeg" /><figcaption>Who is the pound for pound champion of DevOps?</figcaption></figure><p>So its happened…you’re moving to the cloud, its finally a green light to start moving services and applications.</p><p>Your company has gone through the analysis, it makes sense. <strong><em>Now what?</em></strong></p><p>Almost every major company in the world has admitted to themselves the strategic benefit moving to cloud computing can provide, whether its Google Cloud, Azure, or AWS.</p><p>However, senior managers and architects find themselves at many decision junctions in their move to the cloud. Software managers realize that its not just about deciding to use the cloud, but ‘<strong><em>how do we use the cloud</em></strong>?’</p><p>Things like:</p><blockquote>Okay, do we just launch some VM’s?</blockquote><blockquote>Do we containerize?</blockquote><blockquote>Should we adopt microservices patterns?</blockquote><blockquote>If we do, should we use a framework?</blockquote><blockquote>Should we use CloudFormation templates on AWS?</blockquote><blockquote><strong><em>Will that lock us into AWS?</em></strong></blockquote><p>The decision to move to the cloud presents companies with thousands more decisions that can be difficult to navigate, and I wanted to address one of those areas today from my experience as a senior cloud architect: c<strong>ontainers or serverless architecture?</strong></p><h3>Going Serverless: NanoServices At Scale</h3><p>“<em>Welcome to the world of NanoServices, a mystical world where things, are small, petite, and just seem to happen magically on their own!”</em></p><blockquote>But do they?</blockquote><p>The promise of serverless is “no-ops”, meaning you just upload single-logic functions as your backend service endpoints and the cloud provider will take care of the ops for you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EpJzpxyCu8L7NAtNYVqlaA.png" /><figcaption>A simple serverless architecture in AWS</figcaption></figure><p><strong>While serverless does ease the Ops portion of delivering your code, you inherit any upfront advantage back in future technical debt</strong>.</p><p>In other words, as your backend services grow, so does the number of functions. Which means eventually your services start to look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WcVY5PMgndpDrwCPw1N60Q.png" /><figcaption>As your codebase grows, so does the number of functions. Aww, but they’re so tiny and cute!</figcaption></figure><p>I actually experienced this first hand consulting for a FinTech company based out of New York City and Miami. The company had spent upwards of several million dollars developing their backend architecture, of which the complexity of inter-calling Lambda functions made it near impossible to debug anything. Their backend infrastructure became such a ball of yarn and YAML files that untangling the spaghetti pile was near impossible.</p><p><strong>Serverless isn’t all bad though</strong>, there are times where serverless can be a great fit. Scattered, small automation jobs are a perfect fit for serverless architectures, especially when it comes to data pre-processing pipelines. Below is a perfect example:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3RieFBiQjcgWM5oHf928tw.png" /><figcaption>A great use of Cloud Functions in Google Cloud to provide the glue between disparate systems.</figcaption></figure><p>When you need to connect parts of a system together, simple operations that would make deploying a whole microservice seemingly overkill, that’s were going serverless makes sense.</p><blockquote>“Move this to that bucket…”</blockquote><blockquote>“Take that output, send it to this input…”</blockquote><blockquote>“If this, then that…”</blockquote><p>Simple. Simple. And Simple.</p><p>It doesn’t make sense to worry about the ops of glue connectors in cloud architecture, because just like glue, serverless functions have their usage in the cloud toolbox. You wouldn’t glue a whole building together because rivets provide benefits that glue cannot provide, the same goes for serverless architecture.</p><h3>Containers: Packum, Wrapum &amp; Shipum</h3><p>Having dealt with VM’s, serverless, and containers in various scenarios as both a developer and and architecture, I can firmly and confidently say that the future is containerization.</p><p>While containerization requires a additional sets of skills that companies often find their teams lacking (<a href="http://www.kaizentek.io/">we can help</a>), the end results make it well worth the effort.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/495/1*X64Wh7PmNOCThNP5uhUldQ.png" /><figcaption>All your ops in ones beautiful place.</figcaption></figure><p>The reason being is because containers come with a variety of additional advantages, from orchestration frameworks, service meshes, and microservice architectures.</p><p>Containers provide companies with the option to build their backends as <strong>nano or as large</strong> as they would like. You get more freedom with containers, and once they build once, you’re all but guaranteed they’ll build and ship in your pipelines just as they did locally.</p><p>That’s what I truly saw as unique with containers. If it runs well locally, its going to run well in the pipelines and on Kubernetes. <strong>Behavior is more expected and resources are better optimized.</strong></p><p>Once you deploy once, every time after is just automatic with the right CI/CD system in place.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X4YvxtvPoUFNk7Ylk76gnQ.png" /><figcaption>VM’s are very bulk, and require an entire operating system. Container pack your apps and dependencies and leave the OS to the physical machine.</figcaption></figure><p>In general containers offer companies a strategic advantage that serverless cannot deliver.</p><blockquote>Containers provide a cost-savings on resource consumption AND optimal performance, there is no tradeoff between the two.</blockquote><p><strong>You can’t say the same about serverless as an entire backend architecture.</strong></p><p>Yes, while serverless can give you a cost savings (<em>“you only pay for when your function fires!”</em>), but what you get in cost savings, you lose in performance, especially during <strong>function warmup</strong>.</p><p><strong>Function warmup</strong> is when your Lambda or Cloud Function on initial trigger gets moved from a dormant state back into an execution environment by the cloud provider. From my experience, I’ve seen this take as long as 30 seconds, many times over, and that is not an exaggeration.</p><p><strong>In fact, function warmup is such an issue in serverless environments, that whole libraries like </strong><a href="https://github.com/jethrocarr/lambda-ping"><strong>Lambda Ping</strong></a><strong> are used in production just to keep the lambda’s hot and ready.</strong></p><p>In other words, if performance-at-scale is must as your services usage grows, you are going to have some real growing pains in NanoLand.</p><h3>The 12th Round Summary</h3><p>In the bout between serverless and containers, its difficult to clearly say containers is the winner only because I see containers and serverless as two completely different weight classes.</p><blockquote>Containers and serverless shouldn’t be competing in the first place.</blockquote><p>Containers vs. Serverless to me is like saying glue vs. nails, they both have their uses, but don’t get them confused.</p><p>You wouldn’t glue a building together anymore than a woman would get her nails hammered on at the salon. They’re both tools in the cloud toolbox, but knowing when to use each is imperative.</p><h3>The Results:</h3><p><strong>Backend API Architectures</strong><br>Winner: Containers</p><p><strong>Connectors/ Automation Jobs</strong><br>Winner: Serverless</p><h3>Looking For Cloud Architectural Support Or Consulting?</h3><h3>Reach out to us at <a href="http://www.kaizentek.io/">KaizenTek</a>, or call us toll free at (833) 57-CLOUD.</h3><h3>We specialize in cloud architecture, from migrations to machine learning, and have many common deployments packaged and ready to go!</h3><p><a href="https://www.totalwine.com/mobileapp">Check out some of our past work!</a></p><h3>If you liked this post, give me 10 claps and a follow!</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/694/1*rKHq1CrUKuluiCmrjm-TPA.gif" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4288fc999d18" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Magic of Generative Adversarial Network (GANs)]]></title>
            <link>https://becominghuman.ai/the-magic-of-generative-adversarial-network-gans-1c9d5da4a105?source=rss-c21bcd7f76c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/1c9d5da4a105</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-neural-network]]></category>
            <dc:creator><![CDATA[John Anthony Radosta]]></dc:creator>
            <pubDate>Fri, 08 Nov 2019 13:14:00 GMT</pubDate>
            <atom:updated>2019-11-25T14:31:11.600Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/958/1*rBzsscyaphy0NPkScOjYmA.png" /></figure><p>AI is truly some incredible technology. This week I was messing around with building <a href="https://poloclub.github.io/ganlab/">generative adversarial networks</a> (GANs) with TensorFlow and this morning I woke up to some pretty incredible results as my models trained over night.</p><p>In short, GANs are when you have two artificial neural networks that compete against each other, one is a “counterfeiter” and the other is a “detective” (known in <a href="https://becominghuman.ai/">AI</a> parlance as the generator and the discriminator respectively).</p><p>The goal is that the detective network is given a dataset images and told that its job is to learn the dataset, and then discern whether what is fed into it is a fake from the counterfeiter or a real image from the actual dataset. In this case, I used the MNIST dataset which is a dataset of 60,000 images of the number 1–9. Below is an example of one of the images.</p><h3>Trending AI Articles:</h3><blockquote><a href="https://medium.com/@14prakash/back-propagation-is-very-simple-who-made-it-complicated-97b794c97e5c">1. Back-Propagation is very simple. Who made it Complicated ?</a></blockquote><blockquote><a href="https://becominghuman.ai/how-to-train-a-neural-network-to-code-by-itself-a432e8a120df">2. How to train a neural network to code by itself ?</a></blockquote><blockquote><a href="https://becominghuman.ai/introducing-open-mined-decentralised-ai-18017f634a3f">3. Introducing Open Mined: Decentralised AI</a></blockquote><blockquote><a href="https://becominghuman.ai/basics-of-neural-network-bef2ba97d2cf">4. Basics of Neural Network</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hJnwMO4WN-mHeUop" /></figure><p>The counterfeiter is fed complete noise at the start of training and the goal is for the counterfeiter to learn how to change the noise into a picture that will fool the detective network. Below is a picture of the original noise given to the counterfeiter at the start of training.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/892/0*uniUi95Sk-5yjBiQ" /></figure><p>After each training loop, the counterfeiter is given feedback from the detective network as whether if it’s “hotter” or “colder” in terms of its ability to trick the detective network.</p><p>I woke up this morning while both networks trained, and saw how the counterfeiter was progressing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/950/0*-xZyL48yn6_lgYnj" /></figure><p>After about 8.5 hours training on a <a href="https://cloud.google.com/tpu/">TPU in Google Cloud</a>, the counterfeiter is starting to change the noise into what clearly appears to be actual numbers.</p><p>That’s pretty incredible. Imagine learning how to write numbers as a child solely based on your teacher telling you after each attempt “That’s wrong, but you’re getting warmer!”</p><p>You’d probably have given up after 15 minutes. <a href="https://becominghuman.ai/">AI</a> just keeps plowing through. 8.5 hours later, it’s learned how to write numbers correctly from practically nothing.</p><p>This is the future, right here.</p><p><strong>If you enjoyed this post, give me 10 claps and a follow!</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/694/1*lMmW_WjZwZzKjOGuU0975g.gif" /></figure><h3>Don’t forget to give us your 👏 !</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/780/0*2lvCls4yjxVMfZSR" /></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F8f5f8b%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F8f5f8b&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/c43026df6fee7cdb1aab8aaf916125ea/href">https://medium.com/media/c43026df6fee7cdb1aab8aaf916125ea/href</a></iframe><figure><a href="https://becominghuman.ai/artificial-intelligence-communities-c305f28e674c"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*2f7OqE2AJK1KSrhkmD9ZMw.png" /></a></figure><figure><a href="https://upscri.be/8f5f8b"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*v-PpfkSWHbvlWWamSVHHWg.png" /></a></figure><figure><a href="https://becominghuman.ai/write-for-us-48270209de63"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*Wt2auqISiEAOZxJ-I7brDQ.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1c9d5da4a105" width="1" height="1" alt=""><hr><p><a href="https://becominghuman.ai/the-magic-of-generative-adversarial-network-gans-1c9d5da4a105">The Magic of Generative Adversarial Network (GANs)</a> was originally published in <a href="https://becominghuman.ai/">Becoming Human: Artificial Intelligence Magazine</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>