<?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[Dataperformers - Medium]]></title>
        <description><![CDATA[Dataperformers is a world-class AI company made in Montreal. We strive to democratize access to AI by using machine learning to provide practical solutions to real-world problems. - Medium]]></description>
        <link>https://medium.com/dataperformers?source=rss----58d1e43423ff---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Dataperformers - Medium</title>
            <link>https://medium.com/dataperformers?source=rss----58d1e43423ff---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 16:16:43 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/dataperformers" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Affirmative Learning: A new Machine Learning approach toward the AI]]></title>
            <link>https://medium.com/dataperformers/affirmative-learning-a-new-machine-learning-approach-toward-the-ai-d70a428c8bf7?source=rss----58d1e43423ff---4</link>
            <guid isPermaLink="false">https://medium.com/p/d70a428c8bf7</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[iot]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[reinforcement-learning]]></category>
            <dc:creator><![CDATA[Mehdi Merai, Ph.D (c)]]></dc:creator>
            <pubDate>Sun, 22 Dec 2019 05:10:07 GMT</pubDate>
            <atom:updated>2019-12-19T17:22:49.242Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y9bxRfLcsnuj9Ww6TBlxWA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@ryoji__iwata?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Ryoji Iwata</a> on <a href="https://unsplash.com/s/photos/blind?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><blockquote>‘’ We must be careful not to believe things simply because we want them to be true. No one can fool you as easily as you can fool yourself.’’ Richard Feynman.</blockquote><p>A decade ago, teaching a machine to recognize an object or an entity from an image involves a complete “features design” process. This designate a labor work where, in the case of recognizing a human body for example, you should define all the distinctive features that allows to recognize visually an entity human: The body shape, parts, etc. Teaching machines to recognize patterns was a painful process following the complexity of the real world where ML should perform (features complexity, unlimited number of exceptions, etc).</p><p>Then, Deep Learning happen. At the very high level, Deep learning was a major leap in the machine learning and in computer science in general. In fact, this family of approaches tends to automatically recognize features by learning their distribution from a large amount of data. For example, if I would like to teach my machine the ability to recognize cats from any RGB picture, I should expose my deep learning algorithm to a large number of various observations (different pictures of cats) in order to constitute a good internal representation of the entity “cat”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/0*hCgbxGgWLgUnfNYM.gif" /><figcaption>Simplified ConvNet mapping observed features to a learned representation. Photo by <a href="https://upload.wikimedia.org/wikipedia/commons/6/6b/3_filters_in_a_Convolutional_Neural_Network.gif">Wikimedia</a></figcaption></figure><h3>Beyond the Deep Learning</h3><p>After training a machine to recognize a pattern, data scientists and engineers generally follows the following popular and non exclusive avenues to improve their system accuracy:</p><p><strong>(a) Improving the Algorithm itself:</strong> Improving the learning algorithm features representation and learning capabilities: New algorithms, Hyper parameter fine tuning, etc.</p><p><strong>(b) Data enrichment:</strong> Using more and various data to train the learning algorithms.</p><p>We can notice that all of two approaches are improvement setting that are <strong><em>prior</em></strong> to the execution of the Machine Learning system. In fact, the interventions aims to make the AI system ready for a <strong><em>posterior</em></strong> execution in an uncertain and uncontrollable environment. This will suppose that during the execution of the AI system in a real environment, the machine learning system is not set for a short term objective self-adjustment. This is where we identified a room for improvement.</p><h3><strong>Learning algorithms are </strong><strong>freeze into wild exercises</strong></h3><p>We build and train algorithms to recognize pattens mainly in the wild. This phase, most known as the inference phase, consists mainly into mapping real world observations to previously learned representations in order to recognize them.</p><p>To recognize a cat in a camera footage, a trained algorithm will process a specific frame in order scout for learned features that could activate an internal representation of “cat entity”. It’s evident that even a well trained algorithm can still fail to recognize an object simply because there’s an infinite number of exceptional situations that are new from an algorithm perspective: long distance, very particular angle, unforeseen lighting, etc. The real environment is full of uncertainties that make AI naturally fails in uncontrolled conditions, like our world!</p><blockquote>How biologically intelligent agents (like humans) deal with that?</blockquote><h3><strong>Affirmative Learning: The intuition</strong></h3><p>To introduce you the affirmative learning concept, think about the following situation: there is a tiny round metal object sitting on the far extremity of table. It looks like a penny cent or maybe… a metal coat button. What you naturally do to get an affirmation about this? You may came closer, or move your head, you may stand up from your chair to get a better perspective, then you collect a better observation that could have a better probability to fulfill an affirmative answer.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2FTDMbOHni02MZM8fTgS%2Ftwitter%2Fiframe&amp;url=https%3A%2F%2Fgiphy.com%2Fgifs%2Fmib-meninblack-mibinternational-TDMbOHni02MZM8fTgS&amp;image=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2FTDMbOHni02MZM8fTgS%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="234" frameborder="0" scrolling="no"><a href="https://medium.com/media/95861b5c703a878af4e1b9d7a93acedf/href">https://medium.com/media/95861b5c703a878af4e1b9d7a93acedf/href</a></iframe><p>Driven by their affirmation desire, any biologically intelligent agent (humans, some of the mammals, etc) instinctively setup their motricity capabilities by executing a sequence of actions that aims to acquire “a better observation” of the object.</p><p>If we refer to the machine vision area, this could means a controllable acquisition device (a camera for example) that will actively scout for the missing features when it fail to recognize an object with a specific threshold of certainty. The affirmative learning process will be autonomously activated when a <strong><em>predictor algorithm P</em></strong> (like Neural Net based classifier for objects) is able to recognize an object with a low confidence rate.</p><h3>Affirmative Learning: RL driven data acquisition</h3><p>The affirmative learning intuition was inspired (roughly) from the way that biological agents deal with the situation of incertitude. When we desire more certitude, we proceed into an extended data collection.</p><p>Following this sense, our intuition is to take advantage of the capability of the data acquisition device to acquire data, and this driven by a specific (missing) features need. To achieve that, we relay on a reinforcement learning algorithm that we designated as the <strong>Assistant A</strong>. This reinforcement learning agent will control the data acquisition device in order to collect a better observation that allows the Machine Learning algorithm <strong>Predictor P<em> </em></strong>to recognize an entity (or any pattern in general) with a higher confidence.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/972/1*BJZZfFSyPUA5zQNmqd_gTQ.png" /><figcaption><strong>Mehdi Merai Ph.D (c), Ali Elawad Bs.C, @Dataperformers — Patent: US20190171897</strong></figcaption></figure><h3>Affirmative Learning: Use case and market insight</h3><p>Let’s say that we’re controlling a PTZ camera: Pan–Tilt–Optical Zoom. We can map the finite action space <strong><em>(A)</em></strong> of the reinforcement learning based <strong>Assistant A</strong> with the possible discrete controls action that the camera dispose of. This could include more than motor actions like: Contrast, Light, etc. The assistant A will be coupled to a <strong>Predictor P</strong> in order to proceed to a targeted data collection loop on demand following the user set value of the certainty threshold. Trained collect the desired features that the <strong>Predictor P</strong> is missing, the <strong>Assistant A</strong> is trained to collect the features in a specific environment with a minimum number of actions.</p><p>It’s important to notice that affirmative learning is a framework, that designate the interaction mechanics between the <strong>Assistant A</strong> and the <strong>Predictor P</strong>. This means that the framework is agnostic to the specific algorithms of the Predictor or the Assistant. In fact, the nature of the predictor and the assistant could vary following the exercise objectives and the environment of execution.</p><p><strong>Market insight:</strong></p><p>To have a better sense of the market opportunity around this technology, let’s recall the following facts:</p><ul><li>Automation and manufacturing advancements resulted with affordable electronics. Since the price compression, they invaded every consumer part of the market.</li><li>Connectivity democratization pushed the electronic brands to shift from to disconnected things to the Internet of Things.</li><li>Connected devices opened a tremendous market of AI use cases</li></ul><p>We see affirmative vision as raising in the intersection of the mentioned realities. In fact, we see a future where every object could be connected to channel informations and orders. Affirmative vision will be an additional layer of intelligence that will offer to those objects an autonomy regarding the continuous improvement regarding the tasks we will assign to it. A connected camera that have the responsibility to identify cars plates, will have an autonomous way to retrieve from failure by collecting the best possible observation, then by mastering the environment where it exists.</p><p><em>Inventors: </em><a href="https://medium.com/u/ea125d693d48"><em>Mehdi Merai, Ph.D (c)</em></a><em> , </em><a href="https://medium.com/u/d4e26a639244"><em>Ali Elawad</em></a></p><p><em>@Dataperformers</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d70a428c8bf7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/dataperformers/affirmative-learning-a-new-machine-learning-approach-toward-the-ai-d70a428c8bf7">Affirmative Learning: A new Machine Learning approach toward the AI</a> was originally published in <a href="https://medium.com/dataperformers">Dataperformers</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Artificial Intelligence job titles, simplified.]]></title>
            <link>https://medium.com/dataperformers/artificial-intelligence-jobs-simplified-24d7fa3fab60?source=rss----58d1e43423ff---4</link>
            <guid isPermaLink="false">https://medium.com/p/24d7fa3fab60</guid>
            <category><![CDATA[science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[jobs]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Mehdi Merai, Ph.D (c)]]></dc:creator>
            <pubDate>Sat, 21 Sep 2019 15:04:50 GMT</pubDate>
            <atom:updated>2017-09-22T03:16:51.451Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/460/1*SYZ9yWKZKEn8x0ta9RxJCg.jpeg" /></figure><p>Artificial Intelligence is one of the phenomena having the most impact on our lives and economy during the modern technological era. However, an interesting phenomena could be observed; Since it is relatively new, there’s a sort of confusion regarding artificial intelligence skills, jobs and expectations. When the AI abrupt high demand faced the talent shortage reality, a misunderstanding mist has scattered wrong LinkedIn titles and multiple project disappointments.</p><p>Through an example, I will share a simple way that could help both of the talent seekers and the AI providers to communicate better their role and manage their expectations.</p><h4>Machine learning as a Service: the car driver</h4><p>Let’s start with a factual observation. Technically, we don’t need to look for statistics to affirm with a high level of certainty that there are more car drivers than people building or repairing cars. It’s natural.</p><p>Cars are useful. For those who need it, it makes their lives easier and efficient. So in order to drive a car, basically you need some learning and a driving licence. To benefit of its services, you go to your neighbourhood car dealer and depending on your budget, you can acquire a car. Let’s see the situation from an AI perspective. Machine Learning as a service that Google, Amazon or Microsoft provide (the car dealer) is very useful. We need it simply because not all the people are able to build a car. Moreover, not all of them are motivated to understand how a car engine works or have enough of patience to build their own in their backyard.</p><p>In the AI professional jungle, it could be not appropriate to say: “Yes, I do AI” just because you can drive a car. You can even be a cab driver (AI services Integrator), but again, it doesn’t automatically mean that you have the right skills that make you able to assemble a car or design one. Many people are not able to drive, so the cab and Uber drivers area real need in our streets.</p><h4>The AI developer: the car mechanic</h4><p>The car you drive is assembled by some human and many robotic hands. When it breaks, the most evident reaction is to call the tow truck that will take your car to the mechanic. The mechanic (AI developer) is someone who knows how your car works and why it’s not. To be there, he got the right theoretical understanding about engines and has the right knowledge that makes him able to classify the different part of your car. The mechanic or the AI developer understands how each part is interacting with the other. Your mechanic owns some tools (wrench, car lift, diagnostic computer…). He/She should master to make his/her life easy. In the artificial intelligence world, those tools a commonly know as the frameworks like Torch, Tensorflow or Caffe.</p><p>Your mechanic could do an amazing work with your car. He / She could assemble, disassemble your car or even tune its performance. But imagine if I ask him to invent a new kind of Turbo engine. Do you think he /she will agree? I don’t think so. If he / she agrees, I’ll make my distance. If you need someone to design an AI model that solves your untapped problem (there are many of them), you need an AI Scientist.</p><h4>The AI scientist: the car designer</h4><p>How many car brands do exist? Not as much as mechanist I guess. The car designer the person who draws your car, pushes the limits of its engine and thinks your future of your car-related usage. Behind a big sheet of paper and a computer screen, his job is to read, understand, model, try, test, observe and define a concrete approach that improves your state of the art. As a simple driver or passenger, I don’t think that you really need to know who he/she is. You generally don’t have a direct relation with him/her and you even need to master all the details about every line he / she draws.</p><p>Indeed, the car designer knows also how mechanic works. He/She has the right skills that make him / her able to use the different machine learning frameworks in order to test and observe the work. It’s rare to have a job where other people execute and test your models during each trial. It will be long and costly.</p><p>Indeed, the car designer should have a good understanding of the existing models and their limits. He/She needs a deep understanding of the field. You probably understood that the Car designer is rarely a one unique person. I doubt that the person that improved the gas consumption of your car is the same one that worked on the cruise control. It’s almost impossible to find a car designer that masters deeply every tiny aspect about a car. So the car designer is more a specialist job. They have a good understanding of the general mechanisms but they own one or two applications subfield (computer vision, natural language processing…).</p><h4>Why are AI scientists still so highly demanded?</h4><p>In artificial intelligence reality, AI scientists and researchers are still highly demanded for two main reasons:</p><ul><li>Market maturity: AI technology is powerful and transformative. However, it’s fresh. It needs much more time to set up standards and automatic services that fulfill the needs of every industry. The market is full of untapped use cases what could justify the direct implication of an AI scientist.</li><li>AI nature: a learning algorithm owes its intelligence to the data it observes. In reality, your data has a different structure than mine. Apart from rare cases where we share almost the same data structure, semantics and usage, in most of the case, there’s no turnkey learning model ready to perform in my organization. We’re discovering more approaches that could fill this gap (eg: Transfert Learning), but its performance is still not automatic or guaranteed.</li></ul><h4>Owning the car is not enough</h4><p>Owning a car is not enough. You need probably road infrastructures, traffic laws, parking, etc. There’s multiple essential roles roaming around the AI ecosystem without “doing AI”. First, your car needs a proper street (like those in Montreal :) ). Same for your learning model; to perform, it requires a minimum of proper data. This data is made practicable and available thanks to the precious contribution of the road works that we call data workforce (data developer, data scientist, ETL developer, DBA…).</p><p>So, Data scientist, data engineers, back-end developers, user experience designer, … all of them are rarely an option. They are critically important to drive your artificial intelligence project to the success.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=24d7fa3fab60" width="1" height="1" alt=""><hr><p><a href="https://medium.com/dataperformers/artificial-intelligence-jobs-simplified-24d7fa3fab60">Artificial Intelligence job titles, simplified.</a> was originally published in <a href="https://medium.com/dataperformers">Dataperformers</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>