“Artificial Intelligence began with an ancient wish to forge the Gods.”<\/strong><\/em><\/span><\/p>\n<\/p>\n In the singular words of Pamela McCorduck, Artificial Intelligence<\/strong> or AI <\/strong>stems from our desire to “forge the Gods”, a concept that has been around for thousands of years.<\/span><\/p>\n<\/p>\n In the Greek mythology of “The Argonauts and the Quest for the Golden Fleece”<\/strong>, Jason – a mortal hero and leader of the Argonauts – searches for the “Golden Fleece” and along his arduous journey battles Talos—a gargantuan non-human made of solid bronze.<\/span><\/p>\n<\/p>\n Representations of AI also appears in Judaism, Chinese, Japanese, and Indian Philosophy.<\/span><\/p>\n<\/p>\n With the exponential adoption of intelligent voice assistants like Amazon’s Alexa and Apple’s Siri, it appears that AI is a fundamental part of human nature to expand beyond ourselves, an innate quality of humanity’s desire to be Boundless.<\/em><\/span><\/p>\n<\/p>\n With this ancient human dream for both understanding <\/strong>and creating intelligence<\/strong>, let us dive into the wonderful world of Machine Learning <\/strong>or ML.<\/strong><\/span><\/p>\n<\/p>\n Machine Learning as Intelligent Representation<\/strong><\/span><\/p>\n<\/p>\n To truly understand Machine Learning is to truly understand Representation<\/strong>.<\/span><\/p>\n<\/p>\n How we interpret and understand the world as human beings largely depend on our representation of reality<\/strong> (popularized as “mental models”<\/em> by Charlie Munger<\/a>).<\/span><\/p>\n<\/p>\n Take for example the Geocentric <\/strong>Model vs. Heliocentric <\/strong>Model of our own Solar System.<\/span><\/p>\n<\/p>\n As the complicated and nonsensical trajectories of the Geocentric or “Earth-center”<\/em> model clearly shows above, having the correct representation (the Heliocentric or “Sun-center” <\/em>model) makes a world of difference<\/strong>.<\/span><\/p>\n<\/p>\n Likewise, think of Machine Learning as a gateway to the technological holy grail of perfect machine representation of reality <\/strong>– the gateway to the Technological Singularity<\/a>.<\/span><\/p>\n<\/p>\n If you truly understand this (with the preface that knowledge is distinct from understanding), all of the technical jargon and lengthy pitches by technical Founders may start making sense.<\/em><\/span><\/p>\n<\/p>\n Today, I would like to highlight key achievements from decades of Machine Learning Research<\/a>, and how we can apply this understanding <\/strong>to make superior investment decisions for early-stage investing in Machine Learning <\/strong>startup companies.<\/span><\/p>\n<\/p>\n Disclaimer<\/em><\/strong>: even with the current state-of-the-art Deep Learning algorithms, Machine Learning is still in its infancy and really most suitable just for <\/em>Text<\/em><\/strong>, <\/em>Vision<\/em><\/strong>, and <\/em>Voice <\/em><\/strong>applications. This simple 3-Theme Framework will prove invaluable even as you explore the most cutting-edge Machine Learning technologies as it will steer you clear of investment opportunities that are really empty promises like the fraudulent Unicorn <\/em>Theranos<\/em><\/a>.<\/em><\/span><\/p>\n<\/p>\n Text using Recurrent Neural Networks<\/strong><\/span><\/p>\n<\/p>\n We all know and love Google Translate<\/a>, the website that can instantly translate between 100+ different human languages as if by magic.<\/span><\/p>\n<\/p>\n What is the technology behind Google Translate<\/strong>?<\/span><\/p>\n<\/p>\n Source: Radna Intellectual Ventures; Machine Learning for Text Applications.<\/em><\/span><\/p>\n<\/p>\n Without getting into the technical details, the key idea that has made Google Translate possible and the hundreds of other applications for many “AI-powered” Text and Language <\/strong>startups is Recurrent Neural Networks<\/a> or RNNs<\/strong>.<\/span><\/p>\n<\/p>\n Simply stated, RNNs are simplified computer models<\/strong> of the basic building blocks of the human brain—neurons.<\/span><\/p>\n<\/p>\n What makes RNNs smarter (the “recurrent” part) than other “neural networks” is that they continuously improve upon their internal models based on external feedback<\/strong>.<\/span><\/p>\n<\/p>\n This is analogous to how a child learns to walk for the first time<\/a> – the motor neurons work closely with the neurons in the pre-frontal cortex to sense and learn base on feedback<\/strong>.<\/span><\/p>\n<\/p>\n Vision using Convolutional Neural Networks Moving on to Vision<\/strong>, we all have heard about Google’s Waymo<\/a>, General Motors’ Cruise<\/a>, and even Uber<\/a>‘s own efforts in autonomous \/ self-driving cars.<\/span><\/p>\n<\/p>\n What is the key technology behind this revolutionary industry?<\/span><\/p>\n<\/p>\n Source: Radna Intellectual Ventures; Machine Learning for Vision Applications.<\/em><\/span><\/p>\n<\/p>\n Although there are many moving parts (Lidar<\/a>, GPS<\/a>, IMU<\/a>) to any self-driving technology package like Waymo, arguably a central component is the Vision <\/strong>technology.<\/span><\/p>\n<\/p>\n You shouldn’t be surprised to learn that state-of-the-art Computer Vision<\/strong> technology relies heavily on Machine Learning, and specifically a type of “neural network” called Convolutional Neural Networks <\/a>or CNNs<\/strong>.<\/span><\/p>\n<\/p>\n What are CNNs?<\/span><\/p>\n<\/p>\n Think of them as “spatially invariant”<\/em> neural networks.<\/span><\/p>\n<\/p>\n What does that mean?<\/span><\/p>\n<\/p>\n Spatial invariance<\/a> is a simple idea that an object’s identity doesn’t change because of its position in an image or video.<\/span><\/p>\n<\/p>\n With the latest CNNs being a ground-breaking “semantic segmentation”<\/strong> called Gated Shape CNNs<\/a> – we are able to achieve Pixel-level classification<\/strong> of road obstacles.<\/span><\/p>\n<\/p>\n Aside<\/em>: the underlying self-driving tech, including any ML Vision algorithm, simply provide the foundation for solving a higher-level problem known as SLAM<\/a> or simultaneous localization and mapping<\/strong>, the problem being an agent’s ability to simultaneously construct a map of its unknown environment while still keeping track of its location within it.<\/span><\/p>\n<\/p>\n Voice using Transformers<\/strong><\/span><\/p>\n<\/p>\n Alexa<\/em>, let’s conclude with Machine Learning for Voice <\/strong>applications.<\/span><\/p>\n<\/p>\n Transformers <\/strong>– not to be confused with the immense intergalactic robots from the American<\/a> science fiction action films – are the technology behind Natural Language Processing<\/strong>, the heart of everything from Amazon’s Alexa <\/strong>and Apple’s Siri<\/strong>.<\/span><\/p>\n<\/p>\n What are Transformers?<\/span><\/p>\n<\/p>\n Source: Radna Intellectual Ventures; Machine Learning for Voice Applications.<\/em><\/span><\/p>\n<\/p>\n Transformers are a type of Deep Learning<\/strong> architecture <\/strong>first introduced in 2017<\/a> that has exponentially advanced the field of Natural Language Processing<\/strong> or NLP<\/strong>.<\/span><\/p>\n<\/p>\n Within getting into the technical details<\/a>, Transformers consist of a novel computer architecture consisting of a set of “Encoders”<\/strong> chained together with a set of “Decoders.”<\/strong><\/span><\/p>\n<\/p>\n Think of them as a very clever, higher-level systems architecture of the recurrent neural networks<\/em> we mentioned above.<\/span><\/p>\n<\/p>\n Amusing tidbit: Google <\/strong>launched a Transformer project call BERT<\/a> – named after the Muppet character from Sesame Street<\/strong> – that has been trained with huge amounts of voice data and a core piece of the technology behind many Voice-enabled startups.<\/span><\/p>\n<\/p>\n Investable Themes in Machine Learning<\/strong><\/span><\/p>\n<\/p>\n Now that you understand the three key applications of Machine Learning, let’s apply this knowledge to early-stage venture capital investing<\/strong>.<\/span><\/p>\n<\/p>\n Below we map these Machine Learning Themes to Representative Startups.<\/em><\/span><\/p>\n<\/p>\n Source: Radna Intellectual Ventures; Investable Themes (Text, Vision, and Voice) in Machine Learning and Mapping of Machine Learning Themes to Representative Startups.<\/em><\/span><\/p>\n<\/p>\n Of course, you should realize this is not a rigid mapping<\/strong> as there will be companies whose products and services may span and intersect any one or all of these investable themes.<\/span><\/p>\n<\/p>\n For example, consider a Teleoperated Robot-Assisted Surgeon<\/strong>, which would allow a surgeon to operate remotely for a Human-in-the-Loop<\/a> type of system.<\/span><\/p>\n<\/p>\n Amid the COVID-19<\/strong> pandemic, such an AI system may be an attractive investment for say New York-Presbyterian Hospital<\/a> \/ Weill Cornell Medicine<\/a> – the largest hospital system in the U.S. with over 2,000 hospital beds – with its own corporate venture arm NYP Ventures<\/a>.<\/span><\/p>\n<\/p>\n<\/a>© 2020 by Radna Intellectual Ventures, LLC.<\/em><\/span><\/p>\n<\/p>\n
<\/a><\/span><\/p>\n<\/p>\n
<\/a><\/span><\/p>\n
<\/strong><\/span><\/p>\n<\/p>\n<\/a><\/span><\/p>\n
<\/a><\/span><\/p>\n
<\/a><\/span><\/p>\n
\n