<?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 Jordan McDonald on Medium]]></title>
        <description><![CDATA[Stories by Jordan McDonald on Medium]]></description>
        <link>https://medium.com/@jordanmcdonaldmain?source=rss-abdae9593f6e------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*u12HMkROpCKljn3QExk4Nw.jpeg</url>
            <title>Stories by Jordan McDonald on Medium</title>
            <link>https://medium.com/@jordanmcdonaldmain?source=rss-abdae9593f6e------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 01:11:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jordanmcdonaldmain/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[Seeking Simplicity in Machine Learning Project Delivery]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@jordanmcdonaldmain/seeking-simplicity-in-machine-learning-project-delivery-4e2a11ef81b0?source=rss-abdae9593f6e------2"><img src="https://cdn-images-1.medium.com/max/1106/1*GEqgWzVPVNGw6c0FgQi0oQ.png" width="1106"></a></p><p class="medium-feed-snippet">Successfully delivering a Machine Learning project is tough. Seeking simplicity can act as a potent remedy.</p><p class="medium-feed-link"><a href="https://medium.com/@jordanmcdonaldmain/seeking-simplicity-in-machine-learning-project-delivery-4e2a11ef81b0?source=rss-abdae9593f6e------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@jordanmcdonaldmain/seeking-simplicity-in-machine-learning-project-delivery-4e2a11ef81b0?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e2a11ef81b0</guid>
            <category><![CDATA[software]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[project-management]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Tue, 20 Dec 2022 17:26:44 GMT</pubDate>
            <atom:updated>2022-12-20T17:26:44.555Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Rise of Specialism for AI & ML Engineers]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@jordanmcdonaldmain/the-rise-of-specialism-for-ai-ml-engineers-3311dbc4d804?source=rss-abdae9593f6e------2"><img src="https://cdn-images-1.medium.com/max/1908/1*wCD2XYPrPpZUiwQji88oXw.png" width="1908"></a></p><p class="medium-feed-snippet">Is it possible to predict the AI role landscape in five years?</p><p class="medium-feed-link"><a href="https://medium.com/@jordanmcdonaldmain/the-rise-of-specialism-for-ai-ml-engineers-3311dbc4d804?source=rss-abdae9593f6e------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@jordanmcdonaldmain/the-rise-of-specialism-for-ai-ml-engineers-3311dbc4d804?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/3311dbc4d804</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[strategy]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Sun, 24 Apr 2022 16:49:45 GMT</pubDate>
            <atom:updated>2022-04-24T17:25:40.815Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Amplifying Artificial Intelligence Through Internal Conferences]]></title>
            <link>https://medium.com/@jordanmcdonaldmain/amplifying-artificial-intelligence-through-internal-conferences-632ee40f647c?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/632ee40f647c</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[employee-engagement]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[conference]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Fri, 28 Jan 2022 08:51:36 GMT</pubDate>
            <atom:updated>2022-01-28T08:51:36.588Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://www.bazaarvoice.com/">Bazaarvoice</a> has a rich history with Artificial Intelligence (AI) and Machine Learning (ML). AI technology plays a crucial role within our review moderation systems ensuring quality and authenticity, while also enhancing visual content which is part of our <a href="https://www.bazaarvoice.com/products/social-commerce/">Social Commerce</a> offerings and beyond! Despite this, we felt we needed to do more to amplify our use of AI.</p><p>At Bazaarvoice we believe there is substantial customer value to be gained from AI; it can act as a strategic technology lever to drive innovation across the company.</p><p>To translate this vision into action we formed a globally distributed group (including yours truly!) and organised a large-scale internal conference, known as the “AI Summit”.</p><h3>What Is The Bazaarvoice AI Summit?</h3><p>After several months of gruelling (but rewarding!) preparation, in December 2021 we held our inaugural AI Summit. A two day event for <strong>all</strong> of product development to inspire and propel the organisation towards high value AI adoption. While it’s true that in the past (and present!) Bazaarvoice has embraced the technology, the AI Summit was intended to act as jet fuel to accelerate the future use of AI within our products - to further delight our customers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JqpZNE9flvvbVMAPrBjcxg.png" /></figure><p>The Bazaarvoice AI Summit is tied to <strong>two core goals, </strong>both associated with our product development teams:</p><ul><li>Inspire our teams towards <strong>identifying AI opportunities </strong>in their own product.</li><li>Educating our engineering teams on <strong>how to build</strong> AI solutions, setting them on a journey of increased comfort developing AI embedded products.</li></ul><p>Both of these are high impact aspirational goals, but how did we achieve them through the AI Summit?</p><h3>Striving for AI Summit Success</h3><h4>Curated Content</h4><p>We designed the AI Summit to have <strong>something for everyone</strong>. Across the myriad Bazaarvoice products we have hundreds of people who each have varied degrees of technical responsibility. Our mission for the AI Summit was to curate and present content that will engage everyone who supports product development — regardless of their prior experience with AI or how much software development is part of their day to day working life.</p><p>To realise this mission, we collaborated with our partners from Amazon Web Services (AWS) to curate an agenda that included:</p><ul><li>Keynote from CTO Fritz Hesse and Technical Fellow David Anderson that outlined the strategic vision for AI/ML for the organisation.</li><li>Product x AI — A panel discussion from Bazaarvoice product leaders exploring the challenges and opportunity of embedding AI into our products.</li><li>A deep dive into content moderation ML — An exploration of the Machine Learning embedded systems that support the quality and authenticity of Bazaarvoice ratings and reviews</li><li>Much much more!</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*79GASdcDn6XfN_GqSxRL6A.jpeg" /></figure><h4>Engaging The Product Teams</h4><p>The AI Summit <strong>obliterated</strong> the organising teams estimates across all measurable success metrics. This ranged from the volume of attendees, the level of engagement during the sessions, and the enthusiasm of product teams who expressed interest in further exploring how AI can offer value to their clients.</p><p>To compliment our quantified success metrics we gathered qualitative opinions from across our Product Development teams using <a href="https://www.mentimeter.com/">Mentimeter</a>. Some of the feedback we received from our teams is highlighted in the word cloud below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mJT9dPciARy1bs4EZTMVaA.png" /></figure><p>Also, an unexpected number of our happy attendees took to social media to demonstrate their enthusiasm for the AI Summit, sporting our Bazaarvoice “AI Summit” t-shirts!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mZh0C-BpsUfX_vGdOHwAXw.png" /></figure><h3>Looking To The Future</h3><p>The Bazaarvoice AI Summit will have a legacy that goes far beyond the success of the two days in December 2021. It’s a launching pad for revitalising our embrace of AI technologies and how we intent to enhance our products and business offerings over the next several years.</p><p>This is all in service of giving our customers the best possible value across all of Bazaarvoice’s business offerings — Solidifying our position as a market leader while continuing to act as trailblazers who embrace innovative technologies to support our customers needs.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=632ee40f647c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cycling With My Shadow]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@jordanmcdonaldmain/cycling-with-my-shadow-d6639b3bc327?source=rss-abdae9593f6e------2"><img src="https://cdn-images-1.medium.com/max/2191/1*ivnj_4TFC0FtKCzH7pkjBA.jpeg" width="2191"></a></p><p class="medium-feed-link"><a href="https://medium.com/@jordanmcdonaldmain/cycling-with-my-shadow-d6639b3bc327?source=rss-abdae9593f6e------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@jordanmcdonaldmain/cycling-with-my-shadow-d6639b3bc327?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/d6639b3bc327</guid>
            <category><![CDATA[journey]]></category>
            <category><![CDATA[cycling]]></category>
            <category><![CDATA[poetry]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Fri, 11 Sep 2020 10:53:28 GMT</pubDate>
            <atom:updated>2020-09-11T10:57:40.581Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[What I Learned Joining a Company During a Pandemic.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@jordanmcdonaldmain/what-i-learned-joining-a-company-during-a-pandemic-8829a941710a?source=rss-abdae9593f6e------2"><img src="https://cdn-images-1.medium.com/max/1346/1*Yn87PCFcN5EtHPRTqbBq8g.png" width="1346"></a></p><p class="medium-feed-snippet">Imagine you had access to a time machine. First you congratulate yourself for your ingenuity, carefully step into to the craft, then set&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@jordanmcdonaldmain/what-i-learned-joining-a-company-during-a-pandemic-8829a941710a?source=rss-abdae9593f6e------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@jordanmcdonaldmain/what-i-learned-joining-a-company-during-a-pandemic-8829a941710a?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/8829a941710a</guid>
            <category><![CDATA[tips]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[teamwork]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Wed, 24 Jun 2020 12:27:51 GMT</pubDate>
            <atom:updated>2020-06-24T12:27:51.175Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[ONNX — Made Easy.]]></title>
            <link>https://medium.com/data-science/onnx-made-easy-957e60d16e94?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/957e60d16e94</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[onnx]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Sat, 24 Aug 2019 14:23:05 GMT</pubDate>
            <atom:updated>2019-08-25T15:05:47.161Z</atom:updated>
            <content:encoded><![CDATA[<h3>ONNX — Made Easy</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zvUylJb4NkeZn5mq2M54Pw.png" /></figure><p>If you are curious about the varying environmental factors in the current AI technology landscape that have made ONNX a necessity, I have included a snappy addendum at the end of the article. For now, lets directly hone in on the star of the show - <strong>Onnx</strong>.</p><h3>ONNX: So, What Is It Anyway?</h3><p>ONNX is a persistence format. A means to serialise a generated model into a discrete asset that can be stored, versioned and controlled in code. If you peek behind the curtain you will see that ONNX has received significant backing by Microsoft, Facebook, Nvidia &amp; beyond. All organisations who see AI as core to their businesses. With this in mind, you might start thinking, what makes ONNX so special?</p><blockquote><strong>Interoperability</strong>: The ability of different programs to exchange information, share files and use the same protocols.</blockquote><p>This is a simple idea. An Idea that has existed in software engineering for decades (with Java’s JVM a notable example) but has remained absent in the world of AI, a contributing factor in what is known as the ‘replication crisis’ in AI research.</p><p>Essentially ONNX tackles this problem by it giving the guarantee that a model will always remain deterministic in how it behaves when presented with a given data set, with an outcome invariant from the run time environment.</p><p>So, what does this mean for <strong>us</strong>, why should <strong>we </strong>(as data scientists / AI engineers) care?</p><h4><strong>The Pros!</strong></h4><ul><li>You can create a model in one programming language, then execute in an entirely different run time environment (e.g. Imagine creating your model in Python, then deploying onto a C# software solution) — Nice! 😎</li><li>You can create a model using almost any AI tool, the ONNX run time will treat it generically. This yields a generic, consistent interface for any model you create 👏</li><li>With ONNX, you dont need to package the various dependencies used to create your model, all you need is the ONNX runtime ☺️</li><li>It has widespread backing across almost all significant industry contributors. Its open source and its here to stay 😍</li></ul><p>In a Utopian world there would be no issues with ONNX. Despite the frameworks lofty ideals and admirable objectives, there are some wrinkles we need ironed out.</p><h4>The Cons!</h4><ul><li>ONNX is new. Novelty in tech often exposes immaturity in day to day use. For ONNX this manifests as a clunky API and lack of exemplar working implementations in the wild. Production projects rely on certainty and ease of use 😵</li><li>The documentation is limited. Raising the ugly question of ‘what if something goes wrong?’, a frightening risk when the documented support is haphazard 😰</li><li>It directly supports the conversion of a limited volume of libraries into the ONNX format 😓</li><li>There exists (irony in 3…2…1) an absence of examples. This makes adoption a challenge, except for the most rugged and open minded teams 🤯</li></ul><p>In this article, I want to shed a spotlight on the fourth con. Offering a beacon of hope for any would be Goog’ler who encounters the same challenges I did. For each of you ONNX voyagers, I hope these examples will be your lighthouse.</p><h3>ONNX — Made Easy</h3><p>First up, how do we install (this article does not intend to go into any depth on installation, rather to give you a compass point to follow) ONNX on our development environments? Well you need two things.</p><ol><li><a href="https://github.com/onnx/onnx">ONNX</a></li><li><a href="https://github.com/onnx/onnxmltools">ONNXMLTOOLS</a></li></ol><p>Luckily, the setup instructions decompose to a pip installation, instructions are captured in the hyperlinked GitHub repositories. So, you now have ONNX prepared…hmm…so, why are there two dependencies? It really boils down to the decoupling of <em>conversion tools</em> and ONNX itself. Let this handy diagram explain…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d5vrvflEPK0w2TGtZ_uBCg.png" /><figcaption>A tl;dr view on where the ‘ONNXMLTOOLS’ fit into the ONNX eco-system</figcaption></figure><blockquote><strong>Fun Fact:</strong> The conversion process officially supported by ONNX only supports several libraries at the time of writing. For example Sci-Kit Learn has a standalone conversion tool to ONNX (We will see this soon).</blockquote><p>Now that is cleared up (I hope!), lets explore a few code samples that showcase the conversion of a model to ONNX…starting with <strong>XGBoost!</strong></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/36b3d4f9b06a7d94f60e6ec75b45fb47/href">https://medium.com/media/36b3d4f9b06a7d94f60e6ec75b45fb47/href</a></iframe><p>Quite concise, wouldn&#39;t you agree? The example is of course quite trivial but this example shows that the code real estate is quite similar to other persistence methods.</p><p>However, this only considers <em>saving </em>the model, how do we <em>score </em>within the ONNX run time environment?</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/74685306361e345eed902ad4f595dfa0/href">https://medium.com/media/74685306361e345eed902ad4f595dfa0/href</a></iframe><p>This example is once again in Python, however you can use any programming language compatible with the ONNX run time environment. You can imagine this could be quite powerful when integrating a feature driven by an AI to an existing project (with its own pre-established tech stack)!</p><h4>How to Convert ‘Unsupported’ Libraries to ONNX?</h4><p>As we discussed earlier, the current conversion tooling for ONNX supports only a small subset of possible libraries, with a few significant exclusions; Such as <a href="https://scikit-learn.org/stable/">Sci-kit Learn</a>.</p><p>The ONNX open source community has devised a specific library for this purpose (yes… another dependency) dubbed as <a href="https://github.com/onnx/sklearn-onnx">‘sklearn-onnx’</a>. This additional converter is one of several that exist in the ONNX open course ecosystem, with each mirroring the existing standards of the core ONNX tooling (A saving grace).</p><p>The moment of truth - does it have any impact on the code real estate?</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/383c496d3d0b38306149ff2f50ebaa24/href">https://medium.com/media/383c496d3d0b38306149ff2f50ebaa24/href</a></iframe><p>Long story short…no. This is a good outcome, it shows the splintered conversion libraries religiously follow the existing ONNX conversion API — Good.</p><h4>What If I Want to Use Both SKLearn AND XGboost?</h4><p>Good question! This is one area where both prior exemplar and supporting documentation are limited. ONNX, due to sporting several decentralised converters often feels like compiling differing Lego blocks. To enable the SKLearn conversion library to support XGBoost, the objective becomes registering XGBoost from the core ONNX tooling.</p><p>Let’s see what code bootstraps this process…</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/189ff2886ad37d584095a041e15fd682/href">https://medium.com/media/189ff2886ad37d584095a041e15fd682/href</a></iframe><p>Viola! While leveraging each conversion library concurrently <em>feels </em>quite clunky, it works. This demonstrates the base ONNX API as extensible and consistent. If I where to read the tea leaves to predict the future of ONNX, this core principle is one that gives me hope for a more unified tool set in future.</p><p>I Like ONNX. It is attempting to fulfil a real need in the AI community in striving to mandate the sanctity of the models we build. This is essential for reducing the friction of building both AI driven software solutions and software solutions augmented by AI.</p><p>For those in the AI community, we owe it to ourselves to support ONNX in the larvae stage of its development. ONNX is coming. Let’s embrace it.</p><blockquote>So you could be wondering, what got us here? What were the factors that led to the creation of ONNX? I have included a short supplement below that covers my musings on this topic.</blockquote><h3>Cause &amp; Effect — Why Is ONNX a Necessity?</h3><p>The modern AI landscape is a rich, ever changing tapestry of tools &amp; techniques. The past five years have encompassed what I refer to as the modern day AI renaissance, which has manifested itself in three different ways:</p><ul><li>Significant strides made in <em>education</em>, both in the direct vertical of AI and the broader data science skill set.</li><li>AI appreciation has pierced into the day to day lexicon in the world of technology. Its becoming another tool in our belts.</li><li><em>Maturity of tooling;</em> this has established the current ‘three tier’ model of abstraction (e.g. from Cloud cognitive services, <a href="https://aws.amazon.com/sagemaker/">Sagemaker </a>and its ilk to full fat, low level end to end frameworks like <a href="https://www.tensorflow.org/">Tensorflow</a>).</li></ul><p>This article is focused on the consequence of the third point. Tooling has matured in a net positive manner, sure, BUT (yes, there’s a but…) the development of tools has been chaotic and decentralised.</p><blockquote>In just 10 minutes (I timed myself), I counted <strong>24(!)</strong> unique AI focused libraries &amp; platforms. With each offering very similar broad functionality but with frustratingly differing standards, API’s and persistence formats…eugh.</blockquote><p>How did this happen?</p><ul><li>Competitiveness … Big tech has a inherent desire to compete.</li><li>Parity across programming languages for AI does NOT exists in today’s world. Often times functionality is replicated across Python, R, C# &amp; beyond… just in a different library.</li></ul><p>This has a had a very targeted impact on how the models generated by each tools are <em>persisted</em>. Two options have come to light.</p><ol><li>Becoming wholly attached to a proprietary tool / format (or vendor!).</li><li>Using generic serialisation formats like <a href="https://docs.python.org/3/library/pickle.html">Pickle </a>(which have some of their own <a href="https://scikit-learn.org/stable/modules/model_persistence.html#security-maintainability-limitations">concerning </a>aspects.) which alongside JobLib is a common approach to deploying AI models today. I have used this myself on various projects due its ease of use.</li></ol><p>As with all chaotic environments, there exists an opportunity for standardisation and leadership — <strong>In 2017, ONNX entered the fray.</strong></p><p>And the rest, as they say, is history.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=957e60d16e94" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/onnx-made-easy-957e60d16e94">ONNX — Made Easy.</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Doodle to Digital: Extracting Features from Hand Drawn Maps]]></title>
            <link>https://medium.com/kainos-applied-innovation/from-doodle-to-digital-extracting-features-from-hand-drawn-maps-131240e78bea?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/131240e78bea</guid>
            <category><![CDATA[computer-vision]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[maps]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Fri, 25 May 2018 14:45:46 GMT</pubDate>
            <atom:updated>2018-07-30T10:03:08.826Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DTWbavfE3tvYh6FQadHdfw.png" /></figure><p>In our society, technology is trapped in a never-ending state of flux, a forward acceleration thats driven by innovation, widespread access and advancements in hardware. As a consequence businesses are active participants in an ongoing cat and mouse chase attempting to keep up, adapting their business processes to take advantage of new technologies. This transition (from old to new) has a series of challenges, particularly around data that is attached to legacy systems, this is valuable information that needs to be integrated into modern day (digital) processes.</p><p><a href="https://www.kainos.com/">Kainos</a> have vast experience with these challenges, especially in regard to digital transformation for large organisations (including the UK government!). On occasion, customers will approach Kainos with a problem that lives outside the reach of our traditional skill set, problems that often require cutting edge solutions. This is where the innovation capability comes into play, this is an area where the team adds signifanct value to Kainos. Lets dive into an example use case, which involved the innovation team leveraging computer vision to extract features from hand drawn maps for a customer (hereby denoted as ‘CX’).</p><h3><strong>It All Started with a Pencil and a Few Crayons</strong></h3><p>Lets jump back a few decades, back in the old days when all the information associated with an organisation was collected entirely on paper — written manually by humans and stored in multiple, large filing cabinets. Eventually as time went on the digital revolution began to make a big splash, this led to businesses start the slow (often still ongoing!) journey of digitising their own written information for storage on servers.</p><p>CX also found themselves smack bang in the middle of this transition, in the past they captured property information (land ownership etc.) by dispatching a case worker to visit a location (equipped with a pencil and crayons) to draw a birds eye view of the land (while also writing up supporting documentation). The output looked something like this…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8-7pRoZS70T7jabodT9uNg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/836/1*hu_NM75PIroGWiId9ayoJQ.png" /><figcaption>Examples of ‘Hand Drawn’ Maps</figcaption></figure><p>These are whats known as title plans, a scanned drawing which represents a piece of land visually. The anatomy of a title plan is typically made up from colour shading, pencil outlines, north facing arrow and other textual identifiers. Accompanying the visual features associated with title plans are a set of interesting challenges, which wrap a series of constraints around a potential solution.</p><ul><li>Absence of a consistent key between colour to definition, the description of what each shading represents is contained in another distinct document. This means, a case worker manually has to read an entire associated text to determine what each of the coloured areas represent —just imagine how tedious this is! 😔</li><li>The modern digital mapping system contains additional information that is complementary to the hand drawn maps — which means the hand drawn features need to be overlaid in a consistent manner 😬</li><li>Lets not get too down in the dumps…at least the red border is consistent on every title plan 😏</li></ul><p>Okay, so there a few interesting challenges, however we felt that these were barriers we could overcome! So, to summarise, the current business process for a case worker attempting to find all the information for a piece of land looks a little something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ayfAP79uZh2_iRpt5PqXtw.png" /><figcaption>Example work flow for a case worker</figcaption></figure><p>This is not particularly efficient, however this understanding helps define a few core goals that we felt would prove most valuable for CX, including:</p><ol><li>Extract the coloured ‘features’ from the scanned title plans and overlay onto the modern digital maps used by CX.</li><li>Enable easy searching of the associated textual document for the mapping colour key for each title plan.</li></ol><p>Cool, so how we do we actually achieve these goals?</p><h3>Lets Bring Computer Vision to the Party</h3><p>Just incase you have never encountered this concept before, lets start with a handy dandy definition:</p><blockquote>Computer vision is concerned with the automatic extraction, analysis and understanding of useful information from a single image or a sequence of images</blockquote><p>Lets step back, basically what this entails is the extraction of useful information from images — in our case this includes the shaded areas of the scanned map. Within the innovation capablity in Kainos we have extensive experience in computer vision related tasks, using this experience we put together a plan to help achieve our goals, so what did our solution look like?</p><blockquote>Fun Fact: CX posed this challenge to other competitors to see who could solve it most effectively…As you would expect this got our competitive juices flowing.</blockquote><h3>Putting Our Money Where Our Mouth is</h3><p>We chose to put our faith into the grand daddy of computer vision tooling — <a href="https://github.com/opencv/opencv">OpenCV</a>. OpenCV remains one of the best choices when taking on a vision task due to its robustness, integration with multiple languages (we chose Python), community support and the sheer volume of options avaliable which are hidden behind easy to use abstraction layers.</p><ol><li><strong>Image Extraction</strong></li></ol><p>The original data was a scanned PDF containing the hand drawn map &amp; additional supplementary machine text containing details like the title plan ID among other things. To extract the element we wanted (the map) we wrote a script to parse the PDF and output the images contained within it — phew, step one, complete.</p><p><strong>2. Feature Extraction (Aka where the magic happens)</strong></p><p>At this step we had in our possession a dataset containing hand drawn maps, now we had to extract the areas of interest — This is where OpenCV comes into play. We will use the image below as an example to demonstrate how we applied image processing to manipulate the images.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8-7pRoZS70T7jabodT9uNg.png" /><figcaption>The ‘base’ image prior to any image processing</figcaption></figure><p><strong>Step one, Binarization:</strong> We first had to convert the base image into what is known as a binary representation — essentially transforming the image to only show black and white pixels, where white are the ‘features’ of interest and black is everything else.</p><p>However, we encountered an unknown challenge, not all of the hand drawn maps in our dataset had a pure white background — given that these images were essentially scanned from paper we discovered that the paper could be discoloured (taking on a muted yellow appearance). This ruled out the possibility of a brute force strategy which simply removed the black pixels (the pencil lines) and the white pixels (the backgound) — we needed something a little bit more intelligent.</p><p>To overcome this challenge we converted our images to the <a href="https://en.wikipedia.org/wiki/HSL_and_HSV">HSV</a> colour space (rather than the traditional <a href="https://en.wikipedia.org/wiki/RGB_color_space">RGB</a> colour space) which enabled us to target the saturation of the image much more effectively. This enabled us to use the mean saturation of the image as a means of creating a <a href="https://en.wikipedia.org/wiki/Kernel_(image_processing)">mask</a> which could filter out the pixels that are not within the colour ranges where the likes of green, red, bright yellow and blue reside —aka the features we want to extract.</p><p>We then applied this mask to our original image, which yeilded this as the result…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O9V_qkKx6413CdnRmpUjwQ.png" /><figcaption>The intermitent binary image</figcaption></figure><p><strong>Step two, colour retrieval: </strong>We could perform a ‘bitwise and’ operation using the mask &amp; the original image to extract a perfect cut out of the areas of interest, only this time with the colour information intact.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/623/1*n15gR9LDRYbyEMp0sqLSdQ.png" /></figure><p>Success! I think that is pretty good feature extraction 😄</p><p><strong>3. Digital Map Overlay &amp; Association with the Colour Key</strong></p><p>It turns out that overlaying the the hand drawn features on top of the digital mapping system was relatively simple (due to the output image having an alpha channel). Additionally the associated document also contained the coordinate information for the land, which helped us define the location where the overlay would be placed. Great, so this enabled us to achieve the first of our two core goals — blending two distinct sources of information into one central, up to date location.</p><p>However, there is still one big elephant in the room, how could we associate the extracted features with a description stored in a separate (digital) document? Luckily for us the colours leveraged by case workers are quite distinct from one another with very little potential for overlap, with a focus on primary colours, these have each quite different representations in the pixel domain (RGB, where a colour is made up of three brightness values representing the three colour bands).</p><blockquote>Red Range = (255, 180, 180) → (180, 0, 0)</blockquote><blockquote>Green Range = (150, 230, 150) → (25, 100, 25)</blockquote><blockquote>Blue Range = (180, 210, 255) → (0, 70, 180)</blockquote><blockquote>Yellow Range= (255, 255, 150) → (230, 230, 0)</blockquote><p>Using these pixel boundaries we could identify which colours are present in the extracted features. This could then be used as a means of searching the associated document for a key word match, which would enable highlighting the description defining what the colour represents. In one swoop we have removed the need for the case worker to read the entire document for the colour descriptions, enabling them to spend their time on more valuable tasks (while also minimising good old human error).</p><p>So I know what your thinking, what would this look like for a case worker? To showcase this we put together a demo portal which enable a case worker to search for land based on a code, view both the hand drawn and digital features associated with the land as well as extracting the colour key information, check out the screenshot below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G4kP_iBCa_F6wJmLzrHmJw.png" /><figcaption>The basic demo portal we built for CX to showcase the value of our work</figcaption></figure><blockquote>Fun fact #2: It took the core team (three people) only two weeks to create this tool</blockquote><h3>Future Work</h3><p>This particular problem space has a host of additional interesting opportunities that would require computer vision techniques &amp; our innovation team at Kainos sure does love interesting / challenging problems!</p><p>You may have noticed that some of the title plans contain a ‘north’ directional pointer, if we are able to extract this point and adjust for the ‘on paper offset’ it would enable a more accurate overlay of the hand drawn features onto the digital map. So how would we do this? well…</p><ol><li>Train a <a href="https://en.wikipedia.org/wiki/Convolutional_neural_network">convolutional neural network</a> to detect the north arrow in the title plan — extracting bounding box coordinate information.</li></ol><p>Use the orientiation of the north arrow to automate the rotation of the extracted hand drawn features to enable consistency (removing the need for manual editing by a case worker) with the digital map.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/666/1*-ccNaBOQeHekTiuDumxejw.png" /><figcaption>An example of north facing arrow surrounded by the blue box</figcaption></figure><h3>Conclusion</h3><p>Moving data to new systems is hard. Its an ongoing battle that organisations are currently waged in with no end in sight, often this phenemonem breeds some interesting problems, just like this one we faced with CX. Often the solutions are hidden behind interesting technologies like computer vision, where both specific expertise and demonstrable business agility (away from our core Kainos offerings) are given the stage to shine.</p><p>If you feel that anything in this blog is applicable to your own business (even outside of the scope of extracting features from maps), or if you are interested in contacting the innovation team to discover more about what we do, feel free to send us an email: <em>appliedinnovation@kainos.com</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=131240e78bea" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kainos-applied-innovation/from-doodle-to-digital-extracting-features-from-hand-drawn-maps-131240e78bea">From Doodle to Digital: Extracting Features from Hand Drawn Maps</a> was originally published in <a href="https://medium.com/kainos-applied-innovation">Kainos Applied Innovation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Virunga (2014 Documentary) — My Reflections]]></title>
            <link>https://medium.com/@jordanmcdonaldmain/virunga-2014-documentary-my-reflections-bae71d4691ae?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/bae71d4691ae</guid>
            <category><![CDATA[review]]></category>
            <category><![CDATA[documentary]]></category>
            <category><![CDATA[conservation]]></category>
            <category><![CDATA[virunga]]></category>
            <category><![CDATA[life]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Fri, 29 Dec 2017 20:43:10 GMT</pubDate>
            <atom:updated>2017-12-29T20:43:10.177Z</atom:updated>
            <content:encoded><![CDATA[<p>Around two years ago a particular documentary caught my attention, At that time I didn&#39;t commit myself to watching it. Looking back, I wish I saw it sooner. <a href="http://virungamovie.com/"><strong>Virunga</strong></a> had an immediate and profound effect on me, so much so that I feel compelled to write about my thoughts on the documentaries portrayal of the Democratic Republic of the Congo, the Virunga National Park &amp; how the events presented reflect human nature.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gVTenRogkEodX4dtK7WkMA.jpeg" /><figcaption>A Virunga park ranger taking care of a young Mountain Gorilla</figcaption></figure><h3>Surpassed Expectations</h3><p>The film opened with a poignant scene, a funeral of a park ranger. This surprised me as I came into the documentary expecting a core message of conservation, protecting both the environment and the 880 (roughly) remaining Mountain Gorillas in the world — a powerful and important message. However this opening scene set the stage for something bigger, a message further reaching and more important then I realised, something very Human. Virunga tackled not only my original expectations but also murky and complicated topics including:</p><ul><li>Corruption</li><li>War</li><li>Humanitarian Crisis</li></ul><p>There is a special feeling in going into an experience with a preset expectation and having those blown apart; revealing something much more valuable, I will examine a few of the immediate questions Virunga sparked in my mind.</p><h3>The Duality of Human Behaviour</h3><p>As Virunga progresses the landscape portrayed in Congo is revealed to become increasingly dangerous and unstable. Internal conflict continues to escalate with multiple armed groups in direct conflict with the Congo national army while at the same time corporations attempt to take advantage of the countries natural resources (particularly oil). These events showcase the worst side of human behaviour, demonstrating violence, a lust for money &amp; a disregard of right and wrong. This turbulent period in Congo acts as a microcosm for behaviour that has happened for the entirety of human history.</p><blockquote>‘Life is neither good or evil, but only a place for good and evil’ (Marcus Aurelius)</blockquote><p>Despite being caught in the middle of these prominent threats there are those who act as the inverse of these forces by standing both for and protecting something selflessly. The Virunga park rangers. The role of the rangers is to protect the parks environment and wildlife within, they are confronted by numerous challenges. Examples of this we see presented in the documentary include:</p><ul><li>Going undercover and recording instances of bribery from representatives of a corporation (<a href="https://en.wikipedia.org/wiki/SOCO_International">SOCO</a>) hoping to extract oil from Virunga’s Lake Edward.</li><li>Remaining at the national parks headquarters despite being caught between a hammer and an anvil (Congo army and an advancing rebel group) in order to protect what they worked so hard to build.</li></ul><p>Despite this crucible of danger the rangers stand strong and fight for what they believe. An Ideal that yields a resolve greater than the threat of violent tribalism and risks to personal safety by instead showing compassion and determination to protect both the environment and nature.</p><p>One of the rangers André Bauma said something which captures this idea better than I ever could — a simple powerful phrase that speaks volumes and highlights a question we all should strive to answer.</p><blockquote>‘You must justify why you are on this earth — gorillas justify why I am here, they are my life. So if it is about dying, I will die for the gorillas’</blockquote><h3>The Role of the Individual</h3><p>Throughout the documentary the audience was presented the situation through the lens of a few key actors, all of which play a role in protecting the national park. <a href="https://en.wikipedia.org/wiki/Emmanuel_de_Merode">Emmanuel de Merode</a> was one such individual who inspired me, he has been acting director of Virunga National Park since 2008. He stood firm as a direct blocker to SOCO extracting oil from the park while ensuring the rangers remained at their headquarters despite the approaching rebel forces. Emmanuel was the victim of an assassination attempt On April 15, 2014, thankfully he survived and continues to lead the efforts of Virunga National Park.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3lZMVjFwPC2LekGAwtliTg.jpeg" /><figcaption>Emmanuel de Merode</figcaption></figure><p>Behind every ideal that strives for good there exists key individuals who fan the flames of progress and actively drive a positive agenda while acting as a inspirational force. In Virunga where forces for good are surrounded by conflict and strife the role of a leader can not be understated on both an operational level and for outreach; igniting the passion of people just like me to contribute to the common good, if even just a little.</p><h3>The Power of Documentaries</h3><p>Virunga (and the situation in Congo) is one that a lot of people don’t know about, this is why the efforts of journalism as a whole is crucial in today&#39;s society where we are shielded from the issues that plague less fortunate countries. Virunga is an exceptional film, within the first ten minutes I became so engrossed that I started taking notes (a first for me) and by the end of it I (someone who already contributes to the Gorillas in Virunga) felt a need to write about what I saw.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KTJgHbxH3P2vorjMRO92nQ.jpeg" /><figcaption>A Sketch note I drew outlining the key points from the documentary</figcaption></figure><p>This is the power of education. We are in a chaotic world and are often blinded by our ego and <em>importance</em> of our own problems, sometimes it takes a powerful film like Virunga to give someone the kick up the backside they need to contribute to a greater good, if even just a little bit.</p><h3>Conclusion</h3><p>I wrote this article a few hours after watching Virunga, this article represents my immediate impressions on the documentary while advocating the profound effect it had on me. A good documentary will educate, entertain and invoke a sense of empathy. A great documentary will inspire action, reflection on its message and instigate a more informed way of living — I am sure you can guess what category I put Virunga into.</p><p>If you are interested in watching Virunga its currently on Netflix:</p><ul><li><a href="https://www.netflix.com/gb/title/80009431">https://www.netflix.com/gb/title/80009431</a></li></ul><p>If you feel compelled to contribute to the Virunga National Park (I encourage you beforehand to watch the documentary yourself and come to your own conclusions) then refer to the link below:</p><ul><li><a href="https://virunga.org/donate/">https://virunga.org/donate/</a></li></ul><p><strong>Thank you for reading :)</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bae71d4691ae" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Summit 2017: What Things Stood Out?]]></title>
            <link>https://medium.com/@jordanmcdonaldmain/ai-summit-2017-what-things-stood-out-2b726e6918d5?source=rss-abdae9593f6e------2</link>
            <guid isPermaLink="false">https://medium.com/p/2b726e6918d5</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[conference]]></category>
            <dc:creator><![CDATA[Jordan McDonald]]></dc:creator>
            <pubDate>Tue, 23 May 2017 09:20:46 GMT</pubDate>
            <atom:updated>2017-05-23T09:33:58.178Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>AI Summit is a conference and exhibition event which aims to explore the practical implications of leveraging artificial intelligence in business organisations.</em></blockquote><p>To enable this they brought in a wealth of speakers from various organisations that included the likes of Google, Microsoft and others, who are spearheading the abstraction of AI through easy to use platforms. On the other side of the spectrum there were representatives from different sectors such as retail, banking and transport discussing how they are implementing AI into their own businesses.</p><p>In this post I will give an overview of the six key things I took away from the experience, but first take a look at some of the photographs I took while soaking up the vibrant atmosphere at the excellent Business Design Centre in London.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r02NChLJkjnpCiO_E0Ko3Q.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z4vu_CbOqfnT3-ywbFu_ww.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qbt8U4XakxiwbAgXAGq91g.jpeg" /><figcaption>The two talk locations and the exhibition hall</figcaption></figure><h4>AI is at a Tipping Point in Business</h4><p>It was enlightening to see so many companies advocating the use of artificial intelligence (mostly in the sub category of machine learning) in their own businesses with representation in a wide range of industries. This is complimented by the fact that this year’s attendance numbers for the AI Summit broke previous records by a substantial margin, this helps validates the events tagline ‘The 4th industrial revolution is here’.</p><p>This is a compelling marketing line but it needs to be taken with a pinch of salt, the solutions and ideas presented in the summit mostly align with a narrow view of artificial intelligence. This involves using machine learning in particular to solve a problem (for example <a href="https://www.airbnb.com/">Airbnb</a> presented a use case that involves predicting part maintenance) and this is where the value currently lies for businesses.</p><p><strong>Augmentation</strong> over transformation should be the desired thought process for business leaders.</p><h4>Data is King</h4><p>A recurring theme in a large subset of the talks revolved around two problems:</p><ol><li>Acquisition of data</li><li>Cleaning / processing of the data</li></ol><p><a href="https://www.google.pl/url?sa=t&amp;rct=j&amp;q=&amp;esrc=s&amp;source=web&amp;cd=1&amp;cad=rja&amp;uact=8&amp;ved=0ahUKEwiB_NTbpvXTAhViJJoKHdhVCAUQFggiMAA&amp;url=https%3A%2F%2Fwww.crowdflower.com%2F&amp;usg=AFQjCNFDtbIlhwiaX37UlS2XI_1COQJNjQ">CrowdFlower</a> is one company who acknowledged these issues, in particular the importance of having a dataset which is large and varied enough to ensure your machine learning process can generalise with the decision it makes. Machine learning is nothing without data.</p><p>The second point is unsurprising when you consider the eighty / twenty rule in data science, which suggests that 80% of a data scientist’s time will be spent preparing and managing the dataset.</p><p>I spoke with a company called <a href="https://www.tamr.com/">Tamr</a> who have created a service which can simplify this issue by taking a large dataset and transforming it into a format which can utilised more effectively.</p><p>Another company working in the space of data management is <a href="https://www.google.pl/url?sa=t&amp;rct=j&amp;q=&amp;esrc=s&amp;source=web&amp;cd=1&amp;cad=rja&amp;uact=8&amp;ved=0ahUKEwiemYfypvXTAhVGkywKHbxVBdQQFggmMAA&amp;url=https%3A%2F%2Fmty.ai%2F&amp;usg=AFQjCNGh6jEjVEUyeFsyUfnnreUtMOTT_Q">Mighty AI</a> who have proposed a concept around ‘training data as a service’ which can take unstructured, unlabelled data sets which can transform a challenging unsupervised learning problem into something more manageable.</p><h4>Intelligent Agents Will Take Over the World!</h4><p>Over the course of AI Summit I had contact with twenty-nine companies ( one to one discussion or formal talks), fifteen of those companies either focused on, talked about or plan to use intelligent assistants in the future — just over half!</p><p>This is significant and presents a clear indication of where companies particularly in the financial and retail sectors main interests lie in regards to using AI to automate customer interactions.</p><p><a href="https://rainbird.ai/">Rain Bird</a> was one company who showcased their intelligent conversational platform and they demonstrated how it can be used in tandem with a chat bot interface. The platform they provide is context agnostic, can express responses that showcase ‘memory’ and can give rationalisation about its decisions.</p><p>If you are not familiar with chat bots — check out this example:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/478/1*Pzk50N6SlTs61F8s6UE0Ng.gif" /><figcaption>A chat bot which can diagnose symptoms</figcaption></figure><h4>Deep Learning is Entering Businesses</h4><p>Deep learning was a consistent topic that arose during the various talks at AI Summit with a series of valid use cases for this approach, such as those presented by <a href="https://www.google.pl/url?sa=t&amp;rct=j&amp;q=&amp;esrc=s&amp;source=web&amp;cd=1&amp;cad=rja&amp;uact=8&amp;ved=0ahUKEwjEkO6Vp_XTAhWFlSwKHVf1DdoQFggmMAA&amp;url=http%3A%2F%2Fhome.bt.com%2F&amp;usg=AFQjCNFKOrWqP1zPRTU20TOcEeYkSlL-xw">BT</a> which plan to use neural networks to plot the optimum physical network for their hardware systems.</p><p>AI Summit also helped reinforce one thing — deep learning is the current buzz word of machine learning.</p><p>This has its own dangers, while deep learning is a very powerful technique that has multiple potential use cases it is not the right solution for certain problems. Machine learning, like software engineering is about using the right tool for the job and at a few talks I attended they planned to utilise deep learning with insufficient knowledge of their own data and without enough due diligence to consider what other techniques could be more suitable to solve their particular issue.</p><h4>Hardware Designed to Enable AI</h4><p>Machine learning can be time consuming, especially when leveraging deep learning techniques and a <strong>LOT</strong> of data, however there might be a hardware solution to this problem. <a href="https://www.intel.co.uk/content/www/uk/en/homepage.html">Intel</a> gave a talk that revolved around augmenting hardware to tackle this issue by giving an overview of the ‘crest’ line of chips which have been custom made for this purpose.</p><p>Intel’s goal is to challenge the dominance of <a href="http://www.nvidia.co.uk/page/home.html">Nvidia</a> as the de facto machine learning GPU manufacturer by boasting one hundred times reduction in training time by 2020 compared to today’s ‘fastest’ solution.</p><p>My first thought was, this sounds too good to be true? However these will be the first chips created with the intent of optimising one operation:</p><p><strong>Convolution.</strong></p><p>This is the main calculation performed when leveraging convolutional neural networks. An exciting proposition to consider is the ability to ‘rent’ machines using these chips on the cloud, it will be interesting to see to what extent Intel’s claims are validated in the coming years.</p><h4>Importance of Communication in AI</h4><p>AI Summit by its very mission statement as ‘the world’s number one AI event for business’ has a certain audience, business leaders who are not entirely familiar with the complexity involved in AI. This highlights the importance of communicating the value of AI in a digestible manner, that is friendly to a non technical business leader.</p><p>Select talks at AI Summit did not always consider the knowledge of the audience, splattering complex graphs and algorithms onto the screen. Having spoken to a few non technical attendees about some of these talks, a pattern began to emerge with some alarming quotes:</p><blockquote>‘I got lost about half way through the talk’</blockquote><blockquote>‘I am struggling to see the value of artificial intelligence today for us’</blockquote><p>Artificial intelligence is like a rabbit hole filled with complexity. For artificial intelligence to really take off in a business it is important to develop the ability to express difficult concepts in a succinct manner. The business leaders have to ‘get it’ as well.</p><h4>Conclusion</h4><p>AI Summit was an enlightening experience, it helped shed some light on what other companies are doing to integrate artificial intelligence in their own products and services. Overall AI capability across the board is still maturing, even in large corporations. One key lesson from the event is this; now is to time to build capability in AI, get ahead of the curve.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*084ZiZGfAxnXN0cH80ydsg.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2b726e6918d5" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>