POPULARITY
✨ Tutustu Reaktoriin: https://reaktor.comStudiossa professori Risto Miikkulainen. Jakso on kuvattu 7.1.2025.⌚ AIKALEIMAT (0:00) "Terminaattori tulevaisuudesta" (2:01) Star Trek (4:00) Tekoäly (7:04) Hissit (11:12) Optimointi (15:15) Neuroverkot (19:40) Aivot (23:41) Satunnaisuus (26:44) Kuvantunnistus (31:35) Toiminta (36:51) Arkkitehtuuri (44:06) Aktivaatiofunktio (48:15) Häviöfunktio (51:40) Koulutus (54:27) Transformer (1:01:30) Suuret kielimallit (1:10:00) Promptit (1:12:56) "Black box" (1:17:31) Psykedeelit (1:19:22) Skitsofrenia (1:27:10) Autismi (1:30:28) Psykiatria (1:37:11) Kognitio (1:42:38) Muisti (1:45:23) Vapaa tahto (1:47:13) AGI testit (1:53:20) Tietoisuus (1:59:33) Itsetietoisuus (2:09:23) Robotit (2:14:27) Aistit (2:17:25) Tunteet (2:21:38) Eloonjäämisvietti (2:30:33) Liikkuminen (2:35:17) Bologinen laskenta (2:41:08) Elämä (2:49:31) GPU (2:54:40) Evoluutio (3:05:22) Replikaatio (3:11:09) Matrix (3:14:41) Lex Fridman
Today's guest is Risto Miikkulainen, VP of AI Research at Cognizant Advanced AI Labs. Since 2013, the team has focused on research, innovation and the development of advanced AI technologies that further the application of AI to surface opportunities and make better decisions, serving clients and the society at large. Coming together as the Cognizant Advanced AI Lab, the team remains committed to bridging science and the pragmatic application of AI, to foster a better, more productive and innovative future for businesses and society alike. Risto's current research focuses on methods and applications of AI in decision-making, particularly those based on neuroevolution, as well as neural network models of natural language processing and vision. At Cognizant, and previously as a CTO of Sentient Technologies, he is scaling up these approaches to real-world problems. As an AAAI, IEEE and AAIA Fellow, Risto is an author of over 500 articles in these research areas, with his work on neuroevolution recently being recognized with multiple awards. In this episode, Risto talks about: His journey from academia to AI deployment at Cognizant AI Labs, Empowering human decision-making through AI, Use Case: Winning an award for collaboration on improving Botox evaluation, Why AI implementation needs practical adaptation and broad collaboration, An insight into the team's work within research and development, The AI Labs focus on R&D, thought leadership and AI for good, His top advice to require strong fundamentals, critical thinking and adaptability
In this episode, we dive deep into the AI wave and the nexus between academic research and commercial applications. Joining us is Risto Miikkulainen, an esteemed computer science professor at the University of Texas at Austin and AVP of Evolutionary Intelligence at Cognizant Technology Solutions. Together, we explore the journey to the current AI boom, delve into the era of digital natives, and assess Austin's pivotal role in shaping the future of artificial intelligence.Episode HighlightsThe recent explosion of progress in AI has been powered primarily by scaling up existing models with massive amounts of data and compute, rather than fundamental breakthroughs in new techniques.Unlike previous waves of technology adoption, these AI tools were in the hands of consumers before the enterprise, allowing for more fearless experimentation. In a real-world example, Jason's nine-year-old son created nonsensical prompts in Midjourney, yielding fascinating results that facilitated creative thinking and provided unique insights into the AI. The integration of AI into educational platforms like Khan Academy, along with decentralized research tools, raises the intriguing possibility of fostering 'citizen scientists.' However, there is still an open question about how fully this vision could be realized. Risto's current work specializes in using AI to create 'digital twins' of patients, enabling precise simulations to test and optimize various treatment strategies for conditions like stroke and dementia.UT Austin's new online master's programs extend its AI expertise globally while also potentially attracting more talent to Austin and aligning education with local industry needs.What's next Austin? “We have lots of faculty and bright students who are learning about AI. Austin has a thriving startup community that hopefully will grow as a result. And one more factor is the medical school that was started. There are great opportunities, even for these large language models and generative AI in that domain. Try to encourage the AI folks and the medical folks to get together and identify these opportunities. That could be something that becomes a strength in Austin and we can become a world leader in it.”Risto Miikkulainen: Website, LinkedInConstant Therapy Health: WebsiteBoston University Center for Brain Recovery: Website AI Empowers Researchers to Bring Precision Medicine to Post-stroke Speech and Cognitive Rehabilitation Austin Next Links: Website, X/Twitter, YouTube, LinkedIn
Risto Miikkulainen, aka Taso Scofield, is a musician and outcast who has spent his entire life in and around the basketBALL court. He tells Lexman about his passion for the sport and how it has helped him to overcome personal challenges.
Risto Miikkulainen is a doctor and the co-founder of Sympatholytics, a startup that is developing sudaria, a new type of facial moisturizer. In this episode, Lexman and Risto discuss Pasternak and the impact of his work on 20th century literature.
Risto Miikkulainen from Hematology Associates discusses the importance of platelets and capillarities in the elderly. He also discusses frailty and tritheism in relation to this topic. Finally, he talks about Grendel and haematocrit.
WATCH: https://youtu.be/NbGkWjyyLKc Risto Miikkulainen is a Professor of Computer Science at the University of Texas at Austin and AVP of Evolutionary Intelligence at Cognizant AI Labs. He received an M.S. in Engineering from the Helsinki University of Technology (now Aalto University) in 1986, and a Ph.D. in Computer Science from UCLA in 1990. His current research focuses on methods and applications of neuroevolution, evolutionary computation, machine learning, cognitive science as well as neural network models of natural language processing and vision; he is an author of over 450 articles in these research areas. In 2016, he was named Fellow of the Institute of Electrical and Electronics Engineers (IEEE) for contributions to techniques and applications for neural and evolutionary computation. EPISODE LINKS: - Risto's Website https://www.cs.utexas.edu/users/risto/ - Risto's Publications: https://scholar.google.com/citations?user=2SmbjHAAAAAJ&hl=en CONNECT: - Website: https://tevinnaidu.com/podcast - Instagram: https://instagram.com/drtevinnaidu - Facebook: https://facebook.com/drtevinnaidu - Twitter: https://twitter.com/drtevinnaidu - LinkedIn: https://linkedin.com/in/drtevinnaidu TIMESTAMPS: (0:00) - Introduction (1:13) - Evolution of Future Minds (4:37) - Sci-fi Influences (6:29) - Computational Models in Medicine (12:04) - AI & Human Minds (16:45) - Consciousness Models (20:53) - Similarities Between Biology & Computers (24:25) - Computing Human Behaviour (32:18) - Computer Algorithms vs Human Psychology (39:43) - Human Augmentation (Superhuman) (45:11) - Neuroevolution & Evolutionary Computation (1:00:15) - Cognizant AI Labs (1:03:18) - Ethical Dilemmas in AI (1:11:25) - Risto's AI Vision For The Future (1:15:41) - Conclusion Website · YouTube · YouTube
WATCH: https://youtu.be/NbGkWjyyLKc Risto Miikkulainen is a Professor of Computer Science at the University of Texas at Austin and AVP of Evolutionary Intelligence at Cognizant AI Labs. He received an M.S. in Engineering from the Helsinki University of Technology (now Aalto University) in 1986, and a Ph.D. in Computer Science from UCLA in 1990. His current research focuses on methods and applications of neuroevolution, evolutionary computation, machine learning, cognitive science as well as neural network models of natural language processing and vision; he is an author of over 450 articles in these research areas. In 2016, he was named Fellow of the Institute of Electrical and Electronics Engineers (IEEE) for contributions to techniques and applications for neural and evolutionary computation. EPISODE LINKS: - Risto's Website https://www.cs.utexas.edu/users/risto/ - Risto's Publications: https://scholar.google.com/citations?user=2SmbjHAAAAAJ&hl=en CONNECT: - Website: https://tevinnaidu.com/podcast - Instagram: https://instagram.com/drtevinnaidu - Facebook: https://facebook.com/drtevinnaidu - Twitter: https://twitter.com/drtevinnaidu - LinkedIn: https://linkedin.com/in/drtevinnaidu TIMESTAMPS: (0:00) - Introduction (1:13) - Evolution of Future Minds (4:37) - Sci-fi Influences (6:29) - Computational Models in Medicine (12:04) - AI & Human Minds (16:45) - Consciousness Models (20:53) - Similarities Between Biology & Computers (24:25) - Computing Human Behaviour (32:18) - Computer Algorithms vs Human Psychology (39:43) - Human Augmentation (Superhuman) (45:11) - Neuroevolution & Evolutionary Computation (1:00:15) - Cognizant AI Labs (1:03:18) - Ethical Dilemmas in AI (1:11:25) - Risto's AI Vision For The Future (1:15:41) - Conclusion Website · YouTube
Ever wondered how Facebook tracks your ad preferences? Or perhaps what makes your Siri or Alexa tick? Maybe you're simply a philosopher at heart and you enjoy asking questions like "Can machines think?" Inquiries such as these can be explained with the ~simple~ concept of AI! Tune in as we discuss the study and application of AI (i.e. artificial intelligence) in both academic and industry settings with Dr. Risto Miikkulainen, professor from the University of Texas at Austin and associate VP at Cognizant! Decrypting the mysteries of artificial life and decision making, we touch upon the differences between neural networks and evolutionary computation as well as neuroevolution, social change and applied mathematics. They say the future is written in code; you just need to write your way into it! Cognizant's Evolutionary AI Demo: https://evolution.ml/ Check it out!
Futucastin Instagram Kanava: https://www.instagram.com/futucast Yhtenä päivänä Spotifyn fiidiin oli ilmestynyt uusi Lex Fridmanin jakso, jossa vieraana oli mies nimeltä Risto Miikkulainen. Kun olin kuunnellut jakson ja käynyt nopeasti torilla, päätin heti kutsua Riston meillekin vieraaksi. Olin oppinut asioista kuten evoluutiolaskenta, miten elämää voi simuloida tietokoneessa ja se, miten hyeenat oppivat viemään leijonien saaliita. Teille kuuntelijoille piti saada kerrottua näistä asioista saman tien, ja onneksi saimme Riston vieraaksi studioomme. Vielä vuosikymmenien USAssa asumisen jälkeen Risto on poikkeuksellisen hyvä kertoja suomen kielellä. ▶️ Tilaa Youtube kanava: http://www.youtube.com/c/Futucastpodcast?sub_confirmation=1
On this podcast I am joined by Risto Miikkulainen from Cognizant and by Dr. Nuria Oliver from Valencia Spain who led her team to victory in the Cognizant sponsored Xprize Pandemic Response Challenge which was a four month, $500k challenge that focused on the development of data-driven AI systems to predict Covid-19 infection rates and prescribe intervention plans that regional governments, communities and organizations can implement to minimize harm when reopening their economies. Dr. Oliver's team was made up of 20 leading researchers and policy experts from universities and research institutes throughout Valencia, Spain. Click through to see time stamps of the discussion for easier listening. Transcript Time Stamps: 2:16 - Risto tells us about the origins of the Xprize Challenge and why it is so important 4:09 - Dr. Oliver talks about her background and how her work on other pandemics led her to being in a position to put a team together to advice the regional government of Valencia from the beginning of Covid and how that led to entering the Xprize Challenge 8:07 - Dr. Oliver tells us about her team 10:33 - Risto tells us about the different phases of the competition and what challenges competitors faced 12:55 - Dr. Oliver talks about how she and her team approached the competition, how they leveraged their strengths, key strategic approaches and more 18:00 - Dr. Oliver responds to a question I have around whether diversity was a key competitive advantage of her team 21:02 - Risto tells us what he thinks really made Dr. Oliver and her team stand out from the rest of the competition 22:52 - We discuss how this could be applied to other issues and using AI and competitions in general for policy development 32:34 - We discuss the differences, and pluses and minuses, of short competitions vs longer competitions and what is the best approach to big problems
Risto Miikkulainen is a computer scientist at UT Austin. Please support this podcast by checking out our sponsors: – The Jordan Harbinger Show: https://jordanharbinger.com/lex/ – Grammarly: https://grammarly.com/lex to get 20% off premium – Belcampo: https://belcampo.com/lex and use code LEX to get 20% off first order – Indeed: https://indeed.com/lex to get $75 credit EPISODE LINKS: Risto’s Website: https://www.cs.utexas.edu/users/risto/ PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ YouTube Full Episodes: https://youtube.com/lexfridman YouTube Clips: https://youtube.com/lexclips SUPPORT & CONNECT: – Check out the sponsors above, it’s the best way to support this podcast – Support on Patreon: https://www.patreon.com/lexfridman – Twitter:
Computer scientist, Dr Risto Miikkulainen, shows us how we can come up with novel solutions in science by simulating evolution using computers. From bioinformatics to webpage design, the applications of this field are huge.Image with thanks to Helsingin Sanomat https://www.hs.fi/ If you are interested in helping The Biotech Podcast please take 30 seconds to take the following survey: https://harry852843.typeform.com/to/caV6cMzGPaper on surprising anecdotes of evolution: https://direct.mit.edu/artl/article/26/2/274/93255/The-Surprising-Creativity-of-Digital-Evolution-AMicrosite on ESP (Evolutionary Surrogate-Assisted Prescription): https://evolution.ml/esp/Evolutionary Computation software:ECJ: https://cs.gmu.edu/~eclab/projects/ecj/DEAP: https://github.com/DEAP/deap
Evolutionary biologists never have enough time. Some of the most mysterious behaviors in the animal kingdom—like parenting—evolved over thousands of years, if not longer. Human lifespans are just too short to sit and observe such complex behaviors evolve. But computer scientists are beginning to offer clues by using artificial intelligence to simulate the life and death of thousands of generations of animals in a matter of hours or days. It’s called computational evolution. One behavior that’s long baffled biologists is called mobbing, in which a gang of hyenas team up to steal prey from much more powerful lions. When UT Austin computer scientists Risto Miikkulainen and Padmini Rajagopalan simulated hyenas and lions on a virtual African savannah, they found something surprising. Watch a video of real-life hyenas mobbing (courtesy of Michigan State U.): https://youtu.be/Rs7AXFa4sN0 Read more: Evolution of Complex Coordinated Behavior (July 2020) https://www.cs.utexas.edu/users/ai-lab/downloadPublication.php?filename=http://nn.cs.utexas.edu/downloads/papers/rajagopalan.cec2020.pdf&pubid=127822 Music for today’s show was produced by: • Podington Bear - https://www.podingtonbear.com/ • Pogmothoin (a.k.a. Tom Griffin) - https://freesound.org/people/pogmothoin/ Photo credit: Stephanie Dloniak. About Point of Discovery Point of Discovery is a production of the University of Texas at Austin's College of Natural Sciences. You can listen to all our episodes at @point-of-discovery . Questions or comments about this episode, or our series in general? Email Marc Airhart at mairhart[AT]austin.utexas.edu
Evolutionary Artificial Intelligence (AI) technologies are making it possible to discover entirely new objects and behaviors to maximize a given objective and yield solutions that do not yet exist. With the help of massive datasets, machine learning, powerful and distributed computational capacity, Evolutionary AI can help businesses augment decision-making and create new opportunities. In this podcast episode, Jason Stoughton, a veteran technologist, AI enthusiast, author and entrepreneur, speaks with Stephanie Forrest and Risto Miikkulainen, about Evolutionary AI: What it is, how it is being used today, the challenges it faces, the opportunities it offers and what its future might entail. To learn more, visit cognizant.com/ai. The content of this episode is a reproduction of Jason Stoughton’s podcast episode titled ‘Deep Dive Discussion into Evolutionary AI’ that was originally posted on www.thepulseofai.com.
Accurate prediction via machine learning (ML) has always been a challenge for even experienced data scientists. However, AutoML will change the face of ML-based offerings at a core level, suited for all kinds of complex scenarios. In this podcast episode, Jason Stoughton, a veteran technologist, AI enthusiast, author and entrepreneur, speaks to Quoc Le and Risto Miikkulainen on AutoML and the future direction of AI. Risto is an associate VP of Evolutionary AI at Cognizant and also a professor at the University of Texas. Quoc Le needs is one of the preeminent thought leaders in AI today and is known for many things including co-founding Google Brain and as the person behind AutoML. To learn more, visit cognizant.com/ai. The content of this episode is a reproduction of Jason Stoughton’s podcast episode titled ‘AutoML and the Future of AI’ that was originally posted on www.thepulseofai.com.
On this show I am joined by Stephanie Forrest and Risto Miikkulainen to talk about Evolutionary AI: What it is, how it is being used today, the challenges it faces, the opportunities it offers and what its future might entail. If you are at all interested in Evolutionary AI then you will find this discussion extremely informative and interesting. I hope you enjoy it as much as I did.
As businesses and society continue to rely on AI, concerns around trust and explainability have become more apparent. Future adoption of AI can be impacted by a failure to successfully address these issues. In this scenario, Evolutionary AI can be a key tool in helping create more trustworthy AI. In this podcast episode, Jason Stoughton, a veteran technologist, AI enthusiast, author and entrepreneur, speaks with world-renowned AI thought leaders Joydeep Ghosh and Risto Miikkulainen on a wide range of subjects centered around the issue of Trustworthiness in AI and ways that top companies and scientists are successfully addressing this challenge. To learn more, visit cognizant.com/ai. The content of this episode is a reproduction of Jason Stoughton’s podcast episode titled ‘Trustworthiness in AI: Featuring Joydeep Ghosh and Risto Miikkulainen’ that was originally posted on www.thepulseofai.com.
On this podcast I am joined by Quoc Le and Risto Miikkulainen to talk about AutoML and the future direction of AI. Risto is an associate VP of Evolutionary AI at Cognizant and also a professor at the University of Texas. Quoc Le needs little introduction because he is one of the preeminent thought leaders in AI today and is known for many things including co-founding Google Brain and as the person behind AutoML.
World renowned AI thought leader's Joydeep Ghosh and Risto Miikkulainen join me on this podcast in a wide ranging discussion centered around the issue of Trustworthiness in AI. As our businesses and society increasingly rely on AI the issues around trust and explainability have become much more important. And in fact, failure to successfully address the issue will hamper further adoption of AI. Whether you are a business leader, AI scientist, or a policy maker this podcast will help you understand the issues, both technical and philosophical, around explainability and trust in AI, and will give you insights into ways that top companies and scientists are successfully addressing this challenge. We also talk about how Evolutionary AI can be a key tool in helping create more trustworthy AI.
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
Today we’re joined by Risto Miikkulainen, Associate VP of Evolutionary AI at Cognizant AI, and Professor of Computer Science at the UT Austin. Risto joined us back on episode #47 to discuss evolutionary algorithms, and today we do an update of sorts on what is the latest we should know on the topic. In our conversation, we discuss various use cases for evolutionary AI, the relationship between evolutionary algorithms and reinforcement learning, some of the latest approaches to deploying evolutionary models. We also explore his paper “Better Future through AI: Avoiding Pitfalls and Guiding AI Towards its Full Potential,” which details the historical evolution of AI, discussing where things currently stand, and where they might go in the future. The complete show notes for this episode can be found at twimlai.com/talk/367.
This week we interview Risto Miikkulainen, CTO of Sentient AI, to discuss evolutionary computing and the relevance it has for AI. We cover Sentient's work in evolving stock traders to questions on neuroevolution and reinforcement learning. Links: Evolution is the New Deep Learning Experts Weigh in on the Future of AI and Evolutionary Algorithms A Java-Based Evolutionary Computational System The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities An Introduction to Evolutionary Computing Genetic Algorithms with Python Follow us and leave us a rating! iTunes Homepage Twitter @artlyintelly Facebook artificiallyintelligent1@gmail.com
“There's a lot to be discovered if you let evolution run its course and be creative,” says Risto Miikkulainen, CTO of Sentient and professor of computer science at the University of Texas. As the author of over 380 articles covering research topics in neuroevolution and neural network models of language processing and vision, Miikkulainen is an expert in his field. Among many topics, he joins the conversation today to discuss the distinction between neuroevolution and standard neural networks, the novel solutions and unanticipated discoveries that result from evolutionary optimization, the application of neuroevolution to robotics, and Sentient Ascent--a website optimization system built by the team at Sentient, whose job is to optimize the design of websites in order to increase the level of user engagement. Tune in for all the details.
In this week’s episode, Kyle is joined by Risto Miikkulainen, a professor of computer science and neuroscience at the University of Texas at Austin. They talk about evolutionary computation, its applications in deep learning, and how it’s inspired by biology. They also discuss some of the things Sentient Technologies is working on in stock and finances, retail, e-commerce and web design, as well as the technology behind it-- evolutionary algorithms.
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
Today, I'm joined by Kenneth Stanley, Professor in the Department of Computer Science at the University of Central Florida and senior research scientist at Uber AI Labs. Kenneth studied under TWiML Talk #47 guest Risto Miikkulainen at UT Austin, and joined Uber AI Labs after Geometric Intelligence, the company he co-founded with Gary Marcus and others, was acquired in late 2016. Kenneth’s research focus is what he calls Neuroevolution, applies the idea of genetic algorithms to the challenge of evolving neural network architectures. In this conversation, we discuss the Neuroevolution of Augmenting Topologies (or NEAT) paper that Kenneth authored along with Risto, which won the 2017 International Society for Artificial Life’s Award for Outstanding Paper of the Decade 2002 - 2012. We also cover some of the extensions to that approach he’s created since, including, HyperNEAT, which can efficiently evolve very large networks with connectivity patterns that look more like those of the human and that are generally much larger than what prior approaches to neural learning could produce, and novelty search, an approach which unlike most evolutionary algorithms has no defined objective, but rather simply searches for novel behaviors. We also cover concepts like “Complexification” and “Deception”, biology vs computation including differences and similarities, and some of his other work including his book, and NERO, a video game complete with Real-time Neuroevolution. This is a meaty “Nerd Alert” interview that I think you’ll really enjoy.
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
My guest this week is Risto Miikkulainen, professor of computer science at UT-Austin and vice president of Research at Sentient Technologies. Risto came locked and loaded to discuss a topic that we've received a ton of requests for -- evolutionary algorithms. During our talk we discuss some of the things Sentient is working on in the financial services and retail fields, and we dig into the technology behind it, evolutionary algorithms, which is also the focus of Risto’s research at UT. I really enjoyed this interview and learned a ton, and I’m sure you will too! Notes for this show can be found at twimlai.com/talk/47.
In this episode of the SuperDataScience Podcast, I chat with Vice President of Research at Sentient AI, Risto Miikkulainen. You will discuss about the applications of AI across multiple fields, learn about the 2 types of AI - the Evolutionary Algorithms and Reinforcement Learning Algorithms, and also get valuable insights on how AI is changing the employment landscape. If you enjoyed this episode, check out show notes, resources, and more at http://www.superdatascience.com/67
The O'Reilly Radar Podcast: Evolutionary computation, its applications in deep learning, and how it's inspired by biology.In this week’s episode, David Beyer, principal at Amplify Partners, co-founder of Chart.io, and part of the founding team at Patients Know Best, chats with Risto Miikkulainen, professor of computer science and neuroscience at the University of Texas at Austin. They chat about evolutionary computation, its applications in deep learning, and how it’s inspired by biology. Also note, David Beyer's new free report "The Future of Machine Intelligence" is now available for download.Here are some highlights from their conversation: Finding optimal solutions We talk about evolutionary computation as a way of solving problems, discovering solutions that are optimal or as good as possible. In these complex domains like, maybe, simulated multi-legged robots that are walking in challenging conditions—a slippery slope or a field with obstacles—there are probably many different solutions that will work. If you run the evolution multiple times, you probably will discover some different solutions. There are many paths of constructing that same solution. You have a population and you have some solution components discovered here and there, so there are many different ways for evolution to run and discover roughly the same kind of a walk, where you may be using three legs to move forward and one to push you up the slope if it's a slippery slope. You do (relatively) reliably discover the same solutions, but also, if you run it multiple times, you will discover others. This is also a new direction or recent direction in evolutionary computation—that the standard formulation is that you are running a single run of evolution and you try to, in the end, get the optimum. Everything in the population supports finding that optimum. Biological inspiration Some machine learning is simply statistics. It's not simple, obviously, but it is really based on statistics and it's mathematics-based, but some of the inspiration in evolutionary computation and neural networks and reinforcement learning really comes from biology. It doesn't mean that we are trying to systematically replicate what we see in biology. We take the components we understand, or maybe even misunderstand, but we take the components that make sense and put them together into a computational structure. That's what's happening in evolution, too. Some of the core ideas at the very high level of instruction are the same. In particular, there's selection acting on variation. That's the main principle of evolution in biology, and it's also in computation. If you take a little bit more detailed view, we have a population, and everyone is evaluated, and then we select the best ones, and those are the ones that reproduce the most, and we get a new population that's more likely to be better than the previous population. Modeling biology? Not quite yet. There's also developmental processes that most biological systems adapt and learn during their lifetime as well. In humans, the genes specify, really, a very weak starting point. When a baby is born, there's very little behavior that they can perform, but over time, they interact with the environment and that neural network gets set into a system that actually deals with the world. Yes, there's actually some work in trying to incorporate some of these ideas, but that is very difficult. We are very far from actually saying that we really model biology. OSCAR-6 innovates What got us really hooked in this area was that there are these demonstrations where evolution not only optimizes something that you know pretty well, but also comes up with something that's truly novel, something that you don't anticipate. For us, it was this one application where we were evolving a controller for a robot arm, OSCAR-6. It was six degrees of freedom, but you only needed three to really control it. One of the dimensions is that the robot can turn around its vertical axis, the main axis. The goal is to get the fingers of the robot to a particular location in 3D space that's reachable. It's pretty easy to do. We were working on putting obstacles in the way and accidentally disabled the main motor, the one that turns the robot around its main axis. We didn't know it. We ran evolution anyway, and evolution learned and evolved, found a solution that would get the fingers in the goal, but it took five times longer. We only understood what was going on when we put it on screen and looked at the visualization. What the robot was able to do was that when the target was, say, all the way to the left and it needed to turn around the main axis to get the arm close to it, it couldn't do it because it couldn't turn. Instead, it turned the arm from the elbow or shoulder, the other direction, away from the goal, then swung it back real hard; because of inertia, the whole robot would turn around its main axis, even when there was no motor. This was a big surprise. We caused big problems to the robot. We disabled a big, important component of it, but it still found a solution of dealing with it: utilizing inertia, utilizing the physical simulation to get where it needed to go. This is exactly what you would like in a machine learning system. It innovates. It finds things that you did not think about. If you have a robot stuck in a rock in Mars or it loses a wheel, you'd still like it to complete its mission. Using these techniques, we can figure out ways for it to do so.
The O'Reilly Radar Podcast: Evolutionary computation, its applications in deep learning, and how it's inspired by biology.In this week’s episode, David Beyer, principal at Amplify Partners, co-founder of Chart.io, and part of the founding team at Patients Know Best, chats with Risto Miikkulainen, professor of computer science and neuroscience at the University of Texas at Austin. They chat about evolutionary computation, its applications in deep learning, and how it’s inspired by biology. Also note, David Beyer's new free report "The Future of Machine Intelligence" is now available for download.Here are some highlights from their conversation: Finding optimal solutions We talk about evolutionary computation as a way of solving problems, discovering solutions that are optimal or as good as possible. In these complex domains like, maybe, simulated multi-legged robots that are walking in challenging conditions—a slippery slope or a field with obstacles—there are probably many different solutions that will work. If you run the evolution multiple times, you probably will discover some different solutions. There are many paths of constructing that same solution. You have a population and you have some solution components discovered here and there, so there are many different ways for evolution to run and discover roughly the same kind of a walk, where you may be using three legs to move forward and one to push you up the slope if it's a slippery slope. You do (relatively) reliably discover the same solutions, but also, if you run it multiple times, you will discover others. This is also a new direction or recent direction in evolutionary computation—that the standard formulation is that you are running a single run of evolution and you try to, in the end, get the optimum. Everything in the population supports finding that optimum. Biological inspiration Some machine learning is simply statistics. It's not simple, obviously, but it is really based on statistics and it's mathematics-based, but some of the inspiration in evolutionary computation and neural networks and reinforcement learning really comes from biology. It doesn't mean that we are trying to systematically replicate what we see in biology. We take the components we understand, or maybe even misunderstand, but we take the components that make sense and put them together into a computational structure. That's what's happening in evolution, too. Some of the core ideas at the very high level of instruction are the same. In particular, there's selection acting on variation. That's the main principle of evolution in biology, and it's also in computation. If you take a little bit more detailed view, we have a population, and everyone is evaluated, and then we select the best ones, and those are the ones that reproduce the most, and we get a new population that's more likely to be better than the previous population. Modeling biology? Not quite yet. There's also developmental processes that most biological systems adapt and learn during their lifetime as well. In humans, the genes specify, really, a very weak starting point. When a baby is born, there's very little behavior that they can perform, but over time, they interact with the environment and that neural network gets set into a system that actually deals with the world. Yes, there's actually some work in trying to incorporate some of these ideas, but that is very difficult. We are very far from actually saying that we really model biology. OSCAR-6 innovates What got us really hooked in this area was that there are these demonstrations where evolution not only optimizes something that you know pretty well, but also comes up with something that's truly novel, something that you don't anticipate. For us, it was this one application where we were evolving a controller for a robot arm, OSCAR-6. It was six degrees of freedom, but you only needed three to really control it. One of the dimensions is that the robot can turn around its vertical axis, the main axis. The goal is to get the fingers of the robot to a particular location in 3D space that's reachable. It's pretty easy to do. We were working on putting obstacles in the way and accidentally disabled the main motor, the one that turns the robot around its main axis. We didn't know it. We ran evolution anyway, and evolution learned and evolved, found a solution that would get the fingers in the goal, but it took five times longer. We only understood what was going on when we put it on screen and looked at the visualization. What the robot was able to do was that when the target was, say, all the way to the left and it needed to turn around the main axis to get the arm close to it, it couldn't do it because it couldn't turn. Instead, it turned the arm from the elbow or shoulder, the other direction, away from the goal, then swung it back real hard; because of inertia, the whole robot would turn around its main axis, even when there was no motor. This was a big surprise. We caused big problems to the robot. We disabled a big, important component of it, but it still found a solution of dealing with it: utilizing inertia, utilizing the physical simulation to get where it needed to go. This is exactly what you would like in a machine learning system. It innovates. It finds things that you did not think about. If you have a robot stuck in a rock in Mars or it loses a wheel, you'd still like it to complete its mission. Using these techniques, we can figure out ways for it to do so.
In this Halloween episode, Risto Miikkulainen, a Professor of Computer Science and Neuroscience, at the University of Texas at Austin stops by to discuss his work in building machines that are intelligent. He discusses winning the 2K BotPrize for realistic bot game play in Unreal Tournament 2004, what advances in cognitive artificial intelligence means for gaming and training environments, and where we might be going in the future. For more information (including videos), visit www.botprize.org. Show Timeline: • 0:00: Introductions and News of the Week • 12:10: Interview with Risto Mikkulainen • 32:14: Wrap up