<?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 Piotr Godziewski on Medium]]></title>
        <description><![CDATA[Stories by Piotr Godziewski on Medium]]></description>
        <link>https://medium.com/@p.godziewski?source=rss-1cda86cc3891------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*xZvX0LG6BIkGKvOo.</url>
            <title>Stories by Piotr Godziewski on Medium</title>
            <link>https://medium.com/@p.godziewski?source=rss-1cda86cc3891------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 02:59:44 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@p.godziewski/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[Why Is Andrew Ng’s Deep Learning Specialization Unique]]></title>
            <link>https://medium.com/@p.godziewski/why-is-andrew-ngs-deep-learning-specialization-unique-f32dfb9e152f?source=rss-1cda86cc3891------2</link>
            <guid isPermaLink="false">https://medium.com/p/f32dfb9e152f</guid>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Piotr Godziewski]]></dc:creator>
            <pubDate>Fri, 27 Oct 2017 08:24:48 GMT</pubDate>
            <atom:updated>2017-10-31T12:02:44.361Z</atom:updated>
            <content:encoded><![CDATA[<p>I’m part of the analytics platform development and before that worked on several data solutions for internal company use so when I noticed <a href="https://medium.com/@andrewng/deeplearning-ai-announcing-new-deep-learning-courses-on-coursera-43af0a368116">Andrew Ng’s announcement about the brand new AI course</a>, I thought it would be great opportunity to structure my knowledge about the machine learning and specifically its deep learning variant.</p><blockquote>If you wish to verify your own technical understanding of the topic, check out how comfortable you are with <a href="http://playground.tensorflow.org/#activation=tanh&amp;batchSize=10&amp;dataset=circle&amp;regDataset=reg-plane&amp;learningRate=0.03&amp;regularizationRate=0&amp;noise=0&amp;networkShape=4,2&amp;seed=0.60487&amp;showTestData=false&amp;discretize=false&amp;percTrainData=50&amp;x=true&amp;y=true&amp;xTimesY=false&amp;xSquared=false&amp;ySquared=false&amp;cosX=false&amp;sinX=false&amp;cosY=false&amp;sinY=false&amp;collectStats=false&amp;problem=classification&amp;initZero=false&amp;hideText=false">this deep learning emulator</a>.</blockquote><p>If you are fluent with tuning all the parameters (you know exactly what you’re doing and why) then you can probably jump directly to Course 3: Structuring Machine Learning Projects while still checking Course 2: Improving Deep Neural Networks for the techniques you are not familiar with or you want to improve. Otherwise, start from Course 1: Neural Networks and Deep Learning.</p><p>Generally I highly recommend this Specialization for:</p><ul><li>Engineers and specialists being part of cross-functional analytics team involving software development, data science and domain expertise</li><li>Product owners (or product managers) who want to take the data product to the next level but avoid the widespread AI hype and false expectations that may come from it</li><li>Leaders building and developing the analytics team/unit</li><li>AI enthusiasts who are not sure if and how the deep learning can offer real value to achieve the business goals</li></ul><p>As its title and the outline suggests, this is not about:</p><ul><li>Not a high level overview or BI training (be prepared for heavy topics)</li><li>Not a walk-through of particular machine learning framework (however in the last programming exercise the Tensor Flow is introduced for the sake of presenting the benefits of the frameworks)</li><li>Not a deep dive to big data software stack (e.g. Apache Spark)</li></ul><p>You can access <a href="https://www.deeplearning.ai/">deeplearning.ai</a> or go directly to <a href="https://www.coursera.org/specializations/deep-learning">coursera.org</a>. At the time of writing this post, there are three out of five courses available. Below few things which make this course unique in my opinion.​</p><h4><strong>Almost no prerequisite</strong></h4><p>The course outline mentions the prerequisites such as basic programming experience, linear algebra and basic understanding of machine learning purpose. Even if you haven’t touched any code for very long time but you once coded anything and you recall <em>for loops</em>, <em>arrays</em>, <em>vectors</em> and <em>data types</em>, you will not have any problem.</p><p>As for the linear algebra, first few weeks refresh the definition and application of derivatives, loss functions, etc. No aspect or nuance is left unclear.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/859/1*X6CH4TCPIF5byOWPIn8LGw.png" /><figcaption>My favorite nuance; Andrew Ng explains a saddle point :)</figcaption></figure><h4><strong>Machine learning flight simulators</strong></h4><p>A novel approach to walk you through a real-life project and the associated problems. It is a series of what-if questions and decision points testing your knowledge about the methodologies and the best practices.</p><p>I consider this part especially valuable for the leaders. It gives enough insight on the project methodologies so that you can effectively help your team/unit deal with their impediment backlog. Aside from explaining the performance optimization techniques, Andrew Ng spends significant time discussing how you can better structure a project depending on the amount of data at hand.</p><h4><strong>Heroes of Deep Learning</strong></h4><p>This is the optional content in which Andrew Ng interviews the elite researchers and gurus of the deep learning world. Google the names and you realize how much of their research has been productized in every day applications.</p><blockquote>If you need some motivation to start the training, I recommend <a href="https://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html">this story from Google Translator</a>. You’ll read about the involvement of Andrew Ng as well as some other individuals interviewed in Deep Learning courseware.</blockquote><h4><strong>Zero-overhead programming assignments</strong></h4><p>In contrary to many other courses, this one will not require you to install anything locally or create any cloud account. You get access to one of the most popular data science labs, Jupyter Notebooks, right from the Coursera platform. Everything in Python so within reach for the majority of engineers as well as those not actively coding but with the basic programming background.</p><p>You are graded against the course content, i.e. no data parsing, all auxiliary code is clearly commented, works without a problem and lets you realize how much work there is beyond building the deep learning model.</p><h4><strong>Quality + pace = motivation</strong></h4><p>This is top class quality. The pace and the amount of information per training module is just right making it perfect to allocate 45 minutes uninterrupted slot. The programming exercises are well balanced keeping your motivated at the highest level all the time. If you are a cat fan, this course is a must as you start with a cat image recognition model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/710/1*LgKDLjWwRQG1qic-ghxSYw.png" /><figcaption>Cat and non-cat pictures taken around the world, pushed through my cat recognition model</figcaption></figure><h4><strong>Strong brand</strong></h4><p>Andrew Ng is one of the most respected machine learning experts gurus. No doubt that having his course certificate will be a solid point in your personal development plan.</p><h4>Your recommendations?</h4><p>Feel free to share your recommendations in case you develop your own or your team’s machine learning skill set.</p><p><em>EDIT: </em><a href="https://www.coursera.org/learn/data-science-community/home/welcome"><em>The Coursera Data Science Community</em></a><em> is a good place to get more recommendations, discuss the projects and ideas, find peers for discussion.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f32dfb9e152f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Blockchain: Two Learning Paths Before Putting Your Money In]]></title>
            <link>https://medium.com/@p.godziewski/blockchain-two-learning-paths-before-putting-your-money-in-356a4fb27490?source=rss-1cda86cc3891------2</link>
            <guid isPermaLink="false">https://medium.com/p/356a4fb27490</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[investment]]></category>
            <category><![CDATA[bitcoin]]></category>
            <dc:creator><![CDATA[Piotr Godziewski]]></dc:creator>
            <pubDate>Sat, 08 Jul 2017 08:17:23 GMT</pubDate>
            <atom:updated>2017-07-08T08:17:23.529Z</atom:updated>
            <content:encoded><![CDATA[<p>This will not be yet another technical analysis, a touching story of the dotcom bubble or a list of the strongest from the strongest fundamentals behind the altcoins. This is my view on the great learning opportunity which comes with the willingness to add something new to the long term investment portfolio. I personally love learning and typically it is about six months from when I start to consider new instrument/asset till I make a final call about it.</p><p>Learning blockchain is still an existing journey that takes me through the myriad of the technical topics blending with the financial aspects, new business models, security threats, individuals and communities built around their projects.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R9AZ5IDYqHEKrFZEHAWq_A.jpeg" /></figure><h4>Learning Path Option 1: True believers will <a href="https://bitcointalk.org/index.php?topic=375643.0">HODL</a></h4><p>A true believer in the blockchain technology will go for a radical way of investing in it. I call it radical as it is significantly different from what you may know about the traditional financial instruments, valuation techniques, and even trivial aspect of keeping your assets (altcoins and tokens) secured.</p><p>I consider the below list an <strong>absolute minimum </strong>before engaging directly in the crypto market. I believe that the sequence and the references will help you get on the learning curve quicker.</p><ol><li>Distributed ledger technology (DLT) — allows to store the records of data in distributed, decentralized manner while still keeping everything in sync for the users. <em>Distributed </em>reflects multiple peers holding the data (complete records or pieces of it depending on the implementation). <em>Decentralized</em> stands for operations without any central governance such as a cloud provider or a bank. <a href="https://themerkle.com/distributed-ledger-technology-vs-blockchain-technology/">Blockchain is yet another DLT implementation</a> where the new data is added to a block which is then chained to the preceding valid block. What is <em>valid </em>is determined by the system itself based on the given consensus algorithm. How does everything work together? Check out <a href="https://www.youtube.com/watch?v=bBC-nXj3Ng4">this excellent video</a> explaining the principle Bitcoin mechanisms.</li><li>Proof of Work (PoW) vs. Proof of Stake (PoS) and many more — distributed and decentralized system such as blockchain needs to be programmed in order to reach the consensus among the participants. But why to bother so much about underlying technical implementation? The consensus algorithm has significant impact on the system capabilities (e.g. scaling, security and privacy, transaction integrity). I strongly recommend to start with the following <a href="https://assets.kpmg.com/content/dam/kpmg/pdf/2016/06/kpmg-blockchain-consensus-mechanism.pdf">reading from KPMG</a>. Bitcoin runs the Proof of Work. Ethereum has the PoW too but there is a the <a href="https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQ">PoS</a> roadmap (the Casper algorithm). Another important terms that need a proper explanation are mining (in PoW) and forging (often called validating, in PoS). I found <a href="https://ethereum.stackexchange.com/questions/3523/how-do-miners-profit-post-pos-if-at-all-how-will-new-miners-get-their-start-a">this thread</a> perfectly summarizing the impact.</li><li>Bitcoin vs. Ethereum — one can hardly imagine building an altcoin portfolio without BTC (bitcoin) or ETH (ether). You will hear about both competing with each other (especially when the discussion gravitates to the market cap numbers and <a href="https://www.nytimes.com/2017/06/19/business/dealbook/ethereum-bitcoin-digital-currency.html">the flippening</a>) and you will also hear that it <a href="https://www.youtube.com/watch?v=d0x6CtD8iq4">hardly makes sense to compare these two</a> in the context of their features and the corresponding use cases.</li><li>Bitcoin scaling debate — decentralized system, where all decisions lie with the users and there is no central institution controlling it, the computer network incessantly reaching consensus. On the other hand, the bitcoin community of developers, miners and companies has been long struggling with implementing the <a href="https://www.youtube.com/watch?v=DzBAG2Jp4bg">Segregated Witness</a> (SegWit). In short, the bitcoin block will change its structure giving us more transaction capacity. It is not the technical side of SegWit that is the matter of the debate, but the way it is going to be activated (or enforced). Two dominant approaches are <a href="https://www.reddit.com/r/Bitcoin/comments/6hmof7/an_attempt_at_explaining_segwit2x_and_bip_148/">SegWit2x and BIP148</a>. The uncertainty is caused by having <a href="https://bitcoinmagazine.com/articles/countdown-segwit-these-are-dates-keep-eye/">multiple possible scenarios</a> over the next few months. The blockchain split or even some bitcoin lost might be the radical outcomes.</li><li>Vitalik Buterin — as it is usual with emerging tech there are few extremely influential individuals that obviously makes the market quite responsive to the statements they give and the opinions they share as they often have a direct impact on future development. <a href="https://twitter.com/vitalikbuterin">Vitalik Buterin</a> is one of such individuals — the creator and one of the founders of Ethereum. You may want to set news alerts to follow what Buterin has to say about the blockchain future and Ethereum development specifically. If you look for more easy-to-digest learning material, you should refer to <a href="https://antonopoulos.com/">Andreas Antonopoulos</a> and his work. Starting with these two names will lead you to other blockchain gurus (and celebrities).</li><li>Tokens — <a href="https://tokenmarket.net/blockchain/">here</a> is the list of tokens developed on top of various blockchains. No doubt the hottest token sales currently run on top of Ethereum and so to get you on the learning curve quickly you should first familiarize with the concept of <a href="https://github.com/ethereum/wiki/wiki/White-Paper#token-systems">Smart Contracts</a> followed by some reading on the actual <a href="https://www.ethereum.org/token">Ethereum token launch</a> according to its ERC20 standard. This is where things are getting really tricky for non-IT people. It is essential to know the difference between tokens and altcoins as well as tokens and shares/securities, what it takes to launch the token based on crowdfunding or crowdsale and why startups are so eager on this particular way of funding their unfinished products (often not even kicked off). If you don’t fully understand the product vision or you struggle with the idea of tokenizing the product value, you rather stay away from ICO (Initial Coin Offering). Otherwise, your token purchase is speculative. It is not unusual these days that many “<em>investors” </em>blindly buy the tokens without even a look on the product value offering — check out the recently lunched UET (Useless Ethereum Token) and the amount of “<em>investment</em>” it captured.</li></ol><h4>Learning Path Option 2: (Relatively) old-school</h4><p>You’ve done some very limited research about the blockchain technology. You understand the principles of decentralized design of Bitcoin, roughly what mining is about and how it serves the Bitcoin infrastructure. You are comfortable to talk about blockchain use cases beyond peer-to-peer payments such as storage, computing power, prediction system, social networking etc. Yet you don’t feel comfortable speculating in the crypto bubble. Good news.</p><p>There are alternatives allowing you to invest in blockchain without actually buying and trading the digital currency.</p><p>If your risk threshold is more relaxed, search for early adopters and startups which typically focus on the remittance market, development of private enterprise-grade blockchain solutions and integration services. Currently it is unlikely that you find anything on the major exchanges. The selection should be done carefully — quick scan of the companies mentioned <a href="http://investingnews.com/daily/tech-investing/fintech-investing/blockchain-technology-stocks/">here</a> and voilà, there is a <a href="https://www.otcmarkets.com/stock/HSHS/quote">“Buyer Beware” highlight</a> for one of them traded at the OTC Markets (over-the-counter). The same exchange offers the <a href="http://www.otcmarkets.com/stock/GBTC/quote">Bitcoin Investment Trust</a> (GBTC) by <a href="https://grayscale.co/bitcoin-investment-trust/">Grayscale </a>with 1,868,700 shares outstanding, each representing the ownership of 0.09 BTC. Quick math (bitcoin at ~2400USD a coin and GBTC at ~400USD a share when writing this post) means the investors are getting huge premium. Note however that GBTC is not an exchange-traded fund (ETF) lacking the mechanisms for ETF-like accurate price tracking of, in this case, bitcoin.</p><p>Less risky option ,especially if you have been already exposed to the major exchange markets, is to fallback to well-known software and IT players either having blockchain-based offering in their portfolio or being involved in a consortium or open source project. Very few examples to start with:</p><ul><li>Microsoft Azure with its <a href="https://azure.microsoft.com/en-us/solutions/blockchain/">Blockchain-as-a-Service</a> offering supporting Ethereum</li><li><a href="https://www.ibm.com/blockchain/offerings.html">IBM Bluemix</a> with <a href="https://www.hyperledger.org/">Hyperledger </a>blockchain implementation</li><li>GE Predix integration with <a href="https://www.ericsson.com/en/press-releases/2017/5/2107397-ericsson-launches-blockchain-data-integrity-for-ges-predix-platform">Ericsson blockchain-based data integrity service</a></li><li>Joint Accenture and Microsoft effort to create blockchain-based <a href="https://www.reuters.com/article/us-microsoft-accenture-digitalid-idUSKBN19A22B">digital identity network</a></li></ul><p>You may also want to check the members of these two alliances:</p><ul><li><a href="https://entethalliance.org/members/">Enterprise Ethereum Alliance</a></li><li><a href="https://www.hyperledger.org/about/members">Hyperledger</a></li></ul><h4>Next Step: Hands-on</h4><p>Still interested in fueling the blockchain development with your money? If you decide to go the radical way (i.e. trading the altcoins), there is still some hands-on pending before investing any money. By hands-on I mean learning again, particularly the following topics:</p><ul><li>installing and handling your own blockchain wallet(s),</li><li>sending and receiving your first transactions using blockchain,</li><li>knowing the related security threats,</li><li>finding out the tax implications from trading in the crypto space.</li></ul><p><strong><em>Please share this if you find the learning paths valuable. Leave comment about your blockchain learning experience and tips before putting first money in.</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=356a4fb27490" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>