POPULARITY
Can your phone make you a safer driver? Or is it part of the problem? MIT CSAIL Professor Sam Madden shares the origin story of Cambridge Mobile Telematics, a CSAIL spinout using smartphones and AI to prevent car crashes and save lives. In this episode, Professor Madden joins host Kara Miller to explore how accelerometers, algorithms, and well-timed incentives are transforming how we understand and improve driver behavior. Professor Madden reflects on surprising data trends, like why crash rates haven't fallen despite better car tech, and how the ubiquitous smartphone became a key tool for reducing risk on the road. He also dives into how generative AI is reshaping software development, what it means for education and programming skills, and why trust and privacy remain top concerns when deploying AI across sensitive industries. Covering distracted driving, autonomous vehicles, behavioral nudges, and business infrastructure, Professor Madden reveals what's working and where the road ahead leads. Topics include: The history of Cambridge Mobile Telematics The realities of traffic accidents How your phone can track your driving safety Turning your data into a narrative The corrosive effects of phones Learn more about Professor Madden on his website: https://db.csail.mit.edu/madden/ Or his CSAIL page: https://www.csail.mit.edu/person/sam-madden Read our case study on Cambridge Mobile Telematics: https://cap.csail.mit.edu/sites/default/files/2023-05/CMT_Case%20Study%20Template_2022.pdf Connect with CSAIL Alliances: On our site: https://cap.csail.mit.edu/about-us/meet-our-team On X: https://x.com/csail_alliances On LinkedIn: https://linkedin.com/company/mit-csail
Finches, zebras, Darwin... and AI?
Today's episode is a bonus drop from our friends over at the MIT CSAIL Alliances podcast. We'll back in two weeks for Season 11 of Me, Myself, and AI. David Autor, the Daniel (1972) and Gail Rubinfeld Professor, Margaret MacVicar Faculty Fellow in MIT's Department of Economics, says that AI is “not like a calculator where you just punch in the numbers and get the right answer. It's much harder to figure out how to be effective with it.” Offering unique insights into the future of work in an AI-powered world, Autor explains his biggest worries, the greatest upside scenarios, and how he believes we should be approaching AI as a tool, and addresses how AI will impact jobs like nursing and skilled trades. Read the episode transcript here. Studies and papers referenced in this conversation: AI and Product Innovation AI and the Gender Gap Robotics and Nursing Homes CSAIL Alliances connects business and industry to the people and research of MIT's Computer Science and Artificial Intelligence Laboratory. Each month, the CSAIL podcast features cutting-edge MIT and CSAIL experts discussing their current research, challenges, and successes, as well as the potential impact of emerging tech. Follow the podcast here. Me, Myself, and AI is a collaborative podcast from MIT Sloan Management Review and Boston Consulting Group and is hosted by Sam Ransbotham and Shervin Khodabandeh. Our engineer is David Lishansky, and the coordinating producers are Allison Ryder and Alanna Hooper. Stay in touch with us by joining our LinkedIn group, AI for Leaders at mitsmr.com/AIforLeaders or by following Me, Myself, and AI on LinkedIn. We encourage you to rate and review our show. Your comments may be used in Me, Myself, and AI materials.
How AI will Change Your Job with MIT Economics Professor David Autor & The Potential of Self-Supervised Learning with CSAIL PhD Student Sharut Gupta Host: Kara Miller Part One: MIT Economics Professor David Autor says that AI is “not like a calculator where you just punch in the numbers and get the right answer. It's much harder to figure out how to be effective with it.” Offering unique insights into the future of work in an AI-powered world, Professor Autor explains his biggest worries, the greatest upside scenarios, and how he believes we should be approaching AI as a tool, and addresses how AI will impact jobs like nursing and skilled trades. Studies and papers referenced in conversation: AI and Product Innovation: https://aidantr.github.io/files/AI_innovation.p AI and the Gender Gap: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4759218 Robotics and Nursing Homes: https://www.nber.org/papers/w33116 Part Two: CSAIL PhD student Sharut Gutpa describes how self-supervised learning might bring about truly adaptable models which can respond to fast-changing environments, like consumer preferences. CSAIL Alliances connects business and industry to the people and research of MIT's Computer Science and Artificial Intelligence Labs. Learn more about CSAIL Alliances here. Each month, the CSAIL podcast features cutting-edge MIT and CSAIL experts discussing their current research, challenges and successes, as well as the potential impact of emerging tech. Learn more and listen to past episodes. Connect with CSAIL Alliances: On our site (https://cap.csail.mit.edu/about-us/meet-our-team) On X ( / csail_alliances ) On LinkedIn ( / mit-CSAIL )
Daniela Rus, Director of MIT's CSAIL, joins Frazer to explore the intersection of leadership, innovation, and the transformative potential of AI. In this episode, they discuss: How CSAIL has become a global leader in AI and robotics, driving breakthroughs that impact industries worldwide. The essential frameworks for deploying AI ethically and efficiently while building trust in intelligent machines. The role of visionary leadership in using AI to address challenges, inspire innovation, and shape a sustainable, tech-enabled future. — Daniela Rus is the Director of the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT and the Andrew (1956) and Erna Viterbi Professor of Electrical Engineering and Computer Science. A trailblazer in robotics, artificial intelligence, and machine learning, Daniela's groundbreaking research spans soft robotics, underwater exploration, and autonomous systems. Her innovations include printable robots that unfold into functional machines, underwater robots for coral and fish studies, and algorithms for self-driving cars. An IEEE Fellow and member of the National Academy of Engineering, Daniela has earned numerous accolades, including the Engelberger Award for robotics and the IEEE Edison Medal. She holds a PhD in Computer Science from Cornell University.
Have we achieved Artificial General Intelligence? MIT CSAIL Professor Manolis Kellis argues yes. Computers can do nearly every intellectual task that humans are capable of and are rapidly tackling the physical tasks. What does this mean for the future of AI integration, regulation, and development? Hear Professor Kellis' ideas about how businesses can incorporate LLMs (large language models) to minimize silos, why we shouldn't put up too many guardrails on AI technology, and how human-AI collaboration can lead to broader societal benefit, including healthcare. Professor Kellis is working toward a future where AI transforms healthcare through a deeper understanding of our individual genetics and targeted treatment development. Visit Professor Kellis's website to learn more about his work on AI and computational biology: https://www.csail.mit.edu/person/manolis-kellis
This month's podcast is a double feature. First up, Associate Professor and Chief Health AI Officer at the University of California San Diego Karandeep Singh explains the reality of using artificial intelligence for medicine. Professor Singh extrapolates on what works, what doesn't, and how some challenges are social rather than technical. Plus, MIT CSAIL's Assistant Professor Andreea Bobu explains how large language models are advancing the field of robotics. Thanks to LLMs, giving directions to robots might soon look more like a conversation, without the need for step-by-step commands. You can learn more about CSAIL research and access a full transcript of this podcast at https://cap.csail.mit.edu/. To learn about CSAIL's Professional Development Courses, including the upcoming Cybersecurity for Technical Leaders and Deploying AI courses, visit here: cap.csail.mit.edu/events-professional-programs. Podcast listeners save 10% on courses with code MITXPOD10.
Founding Director of the MIT Internet Policy Research Initiative CSAIL Senior Research Scientist Daniel Weitzner says a lack of visibility about how personal data is being used is leading to an erosion of customer trust. However, companies increasingly need to leverage data for analytic advantage, generative AI applications, and more. His research focuses on solutions which would empower consumers with visibility and control of their data, facilitating a future of accountability and trust. Hear more about his work here: https://www.csail.mit.edu/person/daniel-weitzner Find out more about CSAIL Alliances, as well as a full transcript of this podcast, at: https://cap.csail.mit.edu/podcasts/how-companies-can-rebuild-trust-around-data-danny-weitzner?utm_source=alliancesmembers&utm_medium=pardot&utm_campaign=dannypod24 If you would like to learn more about CSAIL's Professional Development Courses, including the upcoming Driving Innovation with Generative AI, visit here: cap.csail.mit.edu/events-professional-programs. Podcast listeners save 10% on courses with code MITXPOD10.
Professor of Finance at the MIT Sloan School of Management and CSAIL Andrew Lo believes AI can help everyday consumers make important financial decisions by democratizing access to quality finance advice. His research aims to address the challenges of deploying AI in finance by, for example, answering questions around responsibility and engaging with financial advisors to make sure such tools are useful in the field. Professor Lo is the faculty director for the FintechAI@CSAIL research initiative. Find out more about CSAIL Alliances, as well as a full transcript of this podcast, at https://cap.csail.mit.edu/podcasts/how-ai-can-help-financial-decision-making-andrew-lo If you would like to learn more about CSAIL's Professional Development Courses, including the upcoming Driving Innovation with Generative AI, visit here: cap.csail.mit.edu/events-professional-programs. Podcast listeners save 10% on courses with code MITXPOD10. Looking for another great podcast? MIT Sloan Management Review's "Me, Myself, and AI," expert hosts and researchers talk with AI leaders from organizations like NASA, Upwork, Github, and Meta to explore how organizations achieve success with generative AI — and what challenges and ethical considerations they face along the way. Listen to Me, Myself, and AI wherever you stream podcasts. https://link.chtbl.com/pxsEZ4pf?sid=CSAIL
Associate Professor Stefanie Mueller, who leads the Human-Computer Interaction group at CSAIL, discusses her groundbreaking research using generative AI for 3D applications. Specifically she explains how generative AI can be combined with mechanical simulation to create stable and personalized 3D models. Find out more about CSAIL Alliances, as well as a full transcript of this podcast, at https://cap.csail.mit.edu. If you would like to learn more about the Cybersecurity for Technical Leaders Course, visit here: https://cap.csail.mit.edu/cybersecurity-technical-leaders-online-course-mit-csail-alliances Podcast listeners save 10% with code MITXTPOD10
MIT Professor Daniel Jackson, associate director of CSAIL and the author of The Essence of Software, argues that if your design is flawed, so is your product. In this podcast, Prof. Jackson shares some cases where software design makes or breaks big companies and what can be done to improve software design in the future. Find out more here: cap.csail.mit.edu
AI has the potential to revolutionize healthcare in areas that range from drug discover to the patient experience. In this podcast, Heather Lane from athenahealth shares the challenges and opportunities of using AI to improve the patient and clinician experience.Heather's Bio:Heather has a PhD from Purdue, where she focused on developing machine learning methods for the computer security problem of anomaly detection. She's worked at the MIT AI Lab (now CSAIL) working with Leslie Kaelbling on reinforcement learning and decision-theoretic planning, Markov decision processes, and the tradeoff between stochastic and deterministic planning.In 2002, she moved to the University of New Mexico as an assistant professor in the Department of Computer Science. There she worked on a number of application areas of ML, including the bioinformatics of RNA interference, genomics, and computational neuroscience (inference of brain activity networks from neuroimaging data). Much of that work involved Bayesian networks and dynamic belief networks.In 2008, she was promoted to associate professor at UNM and was granted tenure. In 2012, she moved from academia to industry, joining Google in Cambridge, MA. working on Knowledge Graph, Google Books, Project Sunroof, and Ads Latency.In 2017, she joined athenahealth to lead a Data Science team working to use athena's immense store of healthcare data to improve healthcare experiences for clinicians and patients.Social LinksYou can follow Heather at: https://www.linkedin.com/in/terranlane/You can follow Maribel at: X/Twitter: https://twitter.com/maribellopezLinkedIn: https://www.linkedin.com/in/maribellopezYouTube: https://www.youtube.com/c/MaribelLopezResearchHashtags: #AI, #Healthcare #PatientExperience
CSAIL's Dr. Neil Thompson joins Kara Miller for a conversation on how AI will affect the future of business. Dr. Thompson also shares insights into the race for computing power and how that competition is shaping the landscape of industry. Find more about Dr. Neil Thompson and his research, as well as a transcript of this podcast at http://cap.csail.mit.edu/podcasts/ais-impact-future-work-neil-thompson.
Die BOLD Community, eine Initiative der Wirtschaftskammer Österreich, ist ein globales Netzwerk von visionären Pionier:innen zur Förderung von Innovation und Zusammenarbeit. Durch Begegnungen von BOLD Minds aus aller Welt entstehen unkonventionelle Ideen und ein reger Wissensaustausch, der Österreich als Innovationshub stärkt. Ein Bold Mind ist Ramin Hasani, ein leitender KI-Wissenschaftler am Computer Science and Artificial Intelligence Lab (CSAIL) des renommierten Massachusetts Institute of Technology (MIT). Zuvor hatte er eine Position als KI- und Machine-Learning--Wissenschaftler bei der Vanguard Group inne und war nebenbei schon als Forschungsmitarbeiter am CSAIL des MIT tätig. Er hat sich in den vergangenen Jahren insbesondere mit Deep Learning-Verfahren und Entscheidungsfindung in komplexen dynamischen Systemen beschäftigt. Seine Doktorarbeit und seine bis heute fortlaufende Forschung zu Liquid Neural Networks wurden international mit zahlreichen Nominierungen und Auszeichnungen, darunter der TÜV Austria Dissertation Award 2020 und der HPC Innovation Excellence Award 2022. Er zählt in diesem Forschungsbereich auch gleichzeitig zu den Pionier:innen. Im Podcast powered by Wirtschaftskammer Österreich sprechen wir mit ihm über seinen faszinierenden Werdegang, warum Liquid Neural Networks in zahlreichen Fachbereichen wie autonome Navigationssysteme, Medizin oder Finanzwelt der Gamechanger schlechthin sein könnten, aber auch welche Unterschiede er als AI-Experte im Approach zwischen den USA und Europa erkennen kann. Am Ende verrät Ramin Hasani außerdem, wann er in seinem Leben BOLD sein musste und wie das seine Karriere geprägt hat. Wenn dir diese Folge gefallen hat, lass uns doch vier, fünf Sterne als Bewertung da und folge dem Podcast auf Spotify, Apple Music und Co. Für Anregungen, Kritik, Feedback oder Wünsche zu künftigen Gästen schick uns jederzeit gerne eine Mail an feedback@trendingtopics.at --- Send in a voice message: https://podcasters.spotify.com/pod/show/trending-topics/message
Die BOLD Community, eine Initiative der Wirtschaftskammer Österreich, ist ein globales Netzwerk von visionären Pionier:innen zur Förderung von Innovation und Zusammenarbeit. Durch Begegnungen von BOLD Minds aus aller Welt entstehen unkonventionelle Ideen und ein reger Wissensaustausch, der Österreich als Innovationshub stärkt. Ein Bold Mind ist Ramin Hasani, ein leitender KI-Wissenschaftler am Computer Science and Artificial Intelligence Lab (CSAIL) des renommierten Massachusetts Institute of Technology (MIT). Zuvor hatte er eine Position als KI- und Machine-Learning--Wissenschaftler bei der Vanguard Group inne und war nebenbei schon als Forschungsmitarbeiter am CSAIL des MIT tätig. Er hat sich in den vergangenen Jahren insbesondere mit Deep Learning-Verfahren und Entscheidungsfindung in komplexen dynamischen Systemen beschäftigt. Seine Doktorarbeit und seine bis heute fortlaufende Forschung zu Liquid Neural Networks wurden international mit zahlreichen Nominierungen und Auszeichnungen, darunter der TÜV Austria Dissertation Award 2020 und der HPC Innovation Excellence Award 2022. Er zählt in diesem Forschungsbereich auch gleichzeitig zu den Pionier:innen. Im Podcast powered by Wirtschaftskammer Österreich sprechen wir mit ihm über seinen faszinierenden Werdegang, warum Liquid Neural Networks in zahlreichen Fachbereichen wie autonome Navigationssysteme, Medizin oder Finanzwelt der Gamechanger schlechthin sein könnten, aber auch welche Unterschiede er als AI-Experte im Approach zwischen den USA und Europa erkennen kann. Am Ende verrät Ramin Hasani außerdem, wann er in seinem Leben BOLD sein musste und wie das seine Karriere geprägt hat. Wenn dir diese Folge gefallen hat, lass uns doch vier, fünf Sterne als Bewertung da und folge dem Podcast auf Spotify, Apple Music und Co. Für Anregungen, Kritik, Feedback oder Wünsche zu künftigen Gästen schick uns jederzeit gerne eine Mail an feedback@trendingtopics.at --- Send in a voice message: https://podcasters.spotify.com/pod/show/trending-topics/message
In episode 85 of The Gradient Podcast, Andrey Kurenkov speaks to Anant AgarwalAnant Agarwal is the chief platform officer of 2U, and founder of edX. Anant taught the first edX course on circuits and electronics from MIT, which drew 155,000 students from 162 countries. He has served as the director of CSAIL, MIT's Computer Science and Artificial Intelligence Laboratory, and is a professor of electrical engineering and computer science at MIT. He is a successful serial entrepreneur, having co-founded several companies including Tilera Corporation, which created the Tile multicore processor, and Virtual Machine Works.Have suggestions for future podcast guests (or other feedback)? Let us know here or reach us at editor@thegradient.pubSubscribe to The Gradient Podcast: Apple Podcasts | Spotify | Pocket Casts | RSSFollow The Gradient on TwitterOutline:* (00:00) Intro* (01:30) History with research* (05:56) Founding EdX* (13:05) AI at EdX* (18:40) Reaction to AI as a teacher* (25:00) Student interest in AI* (32:20) AI's impact on academia* (35:00) Future of AI in education* (38:25) AI writing essays* (43:38) Experiences playing with ChatGPT Get full access to The Gradient at thegradientpub.substack.com/subscribe
CSAIL Professor Sam Hopkins gives an overview of the landscape of high dimensional statistics, and shares some of the research that he and his colleagues at CSAIL are working on to shape it. Professor Hopkins would like to note one correction, when he refers to the "planted clique" problem, it should be "max clique in a random graph". A full transcript of this podcast is available at cap.csail.mit.edu
Taking an inside look at diversity in S.T.E.M., I enjoyed and very open and candid discussion with two ladies offering their insight and perspectives on the possibilities, options, and opportunities in aerospace, science, and technology.Representing are Jackelynne Silva-Martinez, a Systems Engineering Lead and Human Health & Performance Directorate at the NASA Johnson Space Center; and Marlyse Reeves, a PhD student in the Department of Electrical Engineering and Computer Science, CSAIL, MERS Group, at the Massachusetts Institute of Technology (MIT).All Things Aviation & Aerospace is an aviation career webcast live-streamed regularly to provide you insight on the variety of opportunities and possibilities in aviation and aerospace. It's host, Vince Mickens, is a long time private pilot who flies frequently for personal and business. His background includes executive roles with the National Business Aviation Association (NBAA), the Aircraft Owners and Pilots Association (AOPA), and the Bob Hoover Legacy Foundation, all after a 28-year broadcast journalism career working in seven major television and radio broadcast markets nationwide.All Things Aviation & Aerospace is also available on my Private Air Media Group YouTube channel, Facebook Live Page, and Linkedin Profile.https://www.youtube.com/channel/UCnAgvYp8gF4w8WSRdU7Dn4whttps://www.facebook.com/privateairmediagrouphttps://www.linkedin.com/in/vincentmickens-privateairmediagroup-allthingsaviationandaerospace/
Anant Agarwal is the Chief Open Education Officer of 2U/edX. He was the Founder and CEO of edX, an online learning destination founded by Harvard and MIT. Anant taught the first edX course on circuits and electronics from MIT, which drew 155,000 students from 162 countries.He has served as the director of CSAIL, MIT's Computer Science and Artificial Intelligence Laboratory, and is a professor of electrical engineering and computer science at MIT. He is a successful serial entrepreneur, having co-founded several companies including Tilera Corporation, which created the Tile multicore processor, and Virtual Machine Works.Anant won the Maurice Wilkes prize for computer architecture, the Yidan Prize for Education Development, and MIT's Smullin and Jamieson prizes for teaching. He holds a Guinness World Record for the largest microphone array, and is an author of the textbook "Foundations of Analog and Digital Electronic Circuits."His work on Organic Computing was selected by Scientific American as one of 10 World-Changing Ideas in 2011, and he was named in Forbes' list of top 15 education innovators in 2012. Anant is a member of the National Academy of Engineering, a fellow of the American Academy of Arts and Sciences, and a fellow of ACM.He hacks on WebSim, which is an online circuits laboratory in his spare time. Anant holds a Ph.D. from Stanford and a bachelor's from IIT Madras.00:00-02:21-Introduction02:22-05:52-What shaped Anant's Journey05:53-10:15- What did Anant see as a difference between different educational paradigms10:16-14:37- What is the difference between the three modalities of teaching14:38-17:59- What support does a student get in a massive open online course18:00-21:52- How to help a student who is struggling to learn21:53-26:23- The merger of edX and 2U26:24-29:06- Motivating the students29:07-34:42- Affordability34:43-37:34- Are there opportunities to merge online and immersive programs for things that require real work areas37:35-40:39- What inspired Anant to start edX 40:40-43:42- How is Anant doing on his goal of educating a billion people?43:43-45:37- What drives Anant to focus himself on education and what leads him to focus on creating a learning environment?45:38-49:32- Will traditional 4-year colleges will be open to the idea of separate teaching from learning?49:33-53:13- Trends in future education53:14-54:11- ClosingThis episode is brought to you by N2N's Illuminate App, The iPaaS for Higher Education. Learn more at https://illuminateapp.com/web/higher-education/Subscribe and listen to more episodes at IlluminateHigherEducation.comContact Anant Agarwal: https://www.linkedin.com/in/agarwaleduLearn more about 2U: https://2u.com/Learn more about edX: https://www.edx.org/
Anant Agarwal is the Chief Open Education Officer of 2U/edX. He was the Founder and CEO of edX, an online learning destination founded by Harvard and MIT. Anant taught the first edX course on circuits and electronics from MIT, which drew 155,000 students from 162 countries.He has served as the director of CSAIL, MIT's Computer Science and Artificial Intelligence Laboratory, and is a professor of electrical engineering and computer science at MIT. He is a successful serial entrepreneur, having co-founded several companies including Tilera Corporation, which created the Tile multicore processor, and Virtual Machine Works.Anant won the Maurice Wilkes prize for computer architecture, the Yidan Prize for Education Development, and MIT's Smullin and Jamieson prizes for teaching. He holds a Guinness World Record for the largest microphone array, and is an author of the textbook "Foundations of Analog and Digital Electronic Circuits."His work on Organic Computing was selected by Scientific American as one of 10 World-Changing Ideas in 2011, and he was named in Forbes' list of top 15 education innovators in 2012. Anant is a member of the National Academy of Engineering, a fellow of the American Academy of Arts and Sciences, and a fellow of ACM.He hacks on WebSim, which is an online circuits laboratory in his spare time. Anant holds a Ph.D. from Stanford and a bachelor's from IIT Madras.00:00-02:21-Introduction02:22-05:52-What shaped Anant's Journey05:53-10:15- What did Anant see as a difference between different educational paradigms10:16-14:37- What is the difference between the three modalities of teaching14:38-17:59- What support does a student get in a massive open online course18:00-21:52- How to help a student who is struggling to learn21:53-26:23- The merger of edX and 2U26:24-29:06- Motivating the students29:07-34:42- Affordability34:43-37:34- Are there opportunities to merge online and immersive programs for things that require real work areas37:35-40:39- What inspired Anant to start edX 40:40-43:42- How is Anant doing on his goal of educating a billion people?43:43-45:37- What drives Anant to focus himself on education and what leads him to focus on creating a learning environment?45:38-49:32- Will traditional 4-year colleges will be open to the idea of separate teaching from learning?49:33-53:13- Trends in future education53:14-54:11- ClosingThis episode is brought to you by N2N's Illuminate App, The iPaaS for Higher Education. Learn more at https://illuminateapp.com/web/higher-education/Subscribe and listen to more episodes at IlluminateHigherEducation.comContact Anant Agarwal: https://www.linkedin.com/in/agarwaleduLearn more about 2U: https://2u.com/Learn more about edX: https://www.edx.org/
CSAIL's David Clark discusses the differences between misinformation and disinformation, and why the latter is a more serious threat. He also examines possible solutions to the disinformation problem. You can find a transcript for this podcast here: https://cap.csail.mit.edu/sites/default/files/resource-pdfs/David%20Clark%20Podcast%202022.pdf
Andy and Dave discuss the latest in AI news and research, including an update from the DARPA OFFSET (OFFensive Swarm-Enabled Tactics) program, which demonstrated the use of swarms in a field exercise, to include one event that used 130 physical drone platforms along with 30 simulated [0:33]. DARPA's GARD (Guaranteeing AI Robustness against Deception) program has released a toolkit to help AI developers test their models against attacks. Undersecretary of Defense for Research and Engineering, Heidi Shyu, announced DoD's technical priorities, including AI and autonomy, hypersonics, quantum, and others; Shyu expressed a focus on easy-to-use human/machine interfaces [3:35]. The White House AI Initiative Office opened an AI Public Researchers Portal to help connect AI researchers with various federal resources and grant-funding programs [8:44]. A Tesla driver faces felony charges (likely a first) for a fatal crash in which Autopilot was in use, though the criminal charges do not mention the technology [12:23]. In research, MIT's CSAIL publishes (worrisome) research on high scoring convolution neural networks that still achieve high accuracy, even in the absence of “semantically salient features” (such as graying out most of the image); the research also contains a useful list of known image classifier model flaws [18:29]. David Ha and Yujin Tang, at Google Brain in Tokyo, published a white paper surveying recent developments in Collective Intelligence for Deep Learning [19:46]. Roman Garnett makes available a graduate-level book on Bayesian Optimization. And Doug Blackiston returns to chat about the latest discoveries with the Xenobots research and kinematic self-replication [21:54].
Imagine nature inspiring us to create robots of the future! Kevin explores the future with Dr. Daniela Rus, who heads MIT's CSAIL program. Hear about the latest research on everything from AI & ML to Human-Computer Interaction and Computational Biology. Daniela Rus Kevin Scott Behind the Tech with Kevin Scott Discover and listen to other Microsoft podcasts.
Daniel Weitzner is Founding Director of the MIT Internet Policy Research Initiative, Principal Research Scientist at CSAIL, and teaches Internet public policy in MIT's Electrical Engineering and Computer Science Department. Weitzner's research pioneered the development of Accountable Systems to enable computational treatment of legal rules. Weitzner was United States Deputy Chief Technology Officer for Internet Policy in the White House, where he led initiatives on privacy, cybersecurity, copyright, and digital trade policies promoting the free flow of information. He was responsible for the Obama Administration's Consumer Privacy Bill of Rights and the OECD Internet Policymaking Principles. Weitzner has been a leader in Internet public policy from its inception, making fundamental contributions to the successful fight for strong online free expression protection in the United States Supreme Court, and for laws that control government surveillance of email and web browsing data. Weitzner has a law degree from Buffalo Law School, and a B.A. in Philosophy from Swarthmore College. His writings have appeared in Science magazine, the Yale Law Review, Communications of the ACM, the Washington Post, Wired Magazine and Social Research. Weitzner is a founder of the Center for Democracy and Technology, led the World Wide Web Consortium's public policy activities, and was Deputy Policy Director of the Electronic Frontier Foundation.
Armando Solar-Lezama is a Professor at MIT, and the Associate Director & COO of CSAIL. He leads the Computer Assisted Programming Group, focused on program synthesis. Armando's PhD thesis is titled, "Program Synthesis by Sketching", which he completed in 2008 at UC Berkeley. We talk about program synthesis & his work on Sketch, how machine learning's role in program synthesis has evolved over time, and more. - Episode notes: https://cs.nyu.edu/~welleck/episode35.html - Follow the Thesis Review (@thesisreview) and Sean Welleck (@wellecks) on Twitter - Find out more info about the show at https://cs.nyu.edu/~welleck/podcast.html - Support The Thesis Review at www.patreon.com/thesisreview or www.buymeacoffee.com/thesisreview
Ein Medzintechnik-Unternehmen hilft gesunden Menschen bei der Arbeit, im Treibhaus ernten Roboter und in Amsterdam steuern sie Bootstaxis. Neue digitale Produkte erobern Märkte.
Granted, there aren't many cities where self-driving water taxis are a viable option, but Amsterdam may just be one of those places.
Rodney Brooks is a roboticist, former head of CSAIL at MIT, and co-founder of iRobot, Rethink Robotics, and Robust.AI. Please support this podcast by checking out our sponsors: – Paperspace: https://gradient.run/lex to get $15 credit – GiveDirectly: https://givedirectly.org/lex to get gift matched up to $300 – BiOptimizers: http://www.magbreakthrough.com/lex to get 10% off – Four Sigmatic: https://foursigmatic.com/lex and use code LexPod to get up to 60% off – SimpliSafe: https://simplisafe.com/lex and use code LEX to get a free security camera EPISODE LINKS: Rodney's Twitter: https://twitter.com/rodneyabrooks Rodney's Blog: http://rodneybrooks.com/blog/ 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
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
Today we're joined by Daniela Rus, director of CSAIL & Deputy Dean of Research at MIT. In our conversation with Daniela, we explore the history of CSAIL, her role as director of one of the most prestigious computer science labs in the world, how she defines robots, and her take on the current AI for robotics landscape. We also discuss some of her recent research interests including soft robotics, adaptive control in autonomous vehicles, and a mini surgeon robot made with sausage casing(?!). The complete show notes for this episode can be found at twimlai.com/go/515.
Once upon a time, people were computers. It's probably hard to imagine teams of people spending their entire day toiling in large grids of paper, writing numbers and calculating numbers by hand or with mechanical calculators, and then writing more numbers and then repeating that. But that's the way it was before the 1979. The term spreadsheet comes from back when a spread, like a magazine spread, of ledger cells for bookkeeping. There's a great scene in the Netflix show Halston where a new guy is brought in to run the company and he's flying through an electro-mechanical calculator. Halston just shuts the door. Ugh. Imagine doing what we do in a spreadsheet in minutes today by hand. Even really large companies jump over into a spreadsheet to do financial projections today - and with trendlines, tweaking this small variable or that, and even having different algorithms to project the future contents of a cell - the computerized spreadsheet is one of the most valuable business tools ever built. It's that instant change we see when we change one set of numbers and can see the impact down the line. Even with the advent of mainframe computers accounting and finance teams had armies of people who calculated spreadsheets by hand, building complicated financial projections. If the formulas changed then it could take days or weeks to re-calculate and update every cell in a workbook. People didn't experiment with formulas. Computers up to this point had been able to calculate changes and provided all the formulas were accurate could output results onto punch cards or printers. But the cost had been in the millions before Digital Equipment and Data Nova came along and had dropped into the tens or hundreds of thousands of dollars The first computerized spreadsheets weren't instant. Richard Mattessich developed an electronic, batch spreadsheet in 1961. He'd go on to write a book called “Simulation of the Firm Through a Budget Computer Program.” His work was more theoretical in nature, but IBM developed the Business Computer Language, or BCL the next year. What IBM did got copied by their seven dwarves. former GE employees Leroy Ellison, Harry Cantrell, and Russell Edwards developed AutoPlan/AutoTab, another scripting language for spreadsheets, following along delimited files of numbers. And in 1970 we got LANPAR which opened up more than reading files in from sequential, delimited sources. But then everything began to change. Harvard student Dan Bricklin graduated from MIT and went to work for Digital Equipment Corporation to work on an early word processor called WPS-8. We were now in the age of interactive computing on minicomputers. He then went to work for FasFax in 1976 for a year, getting exposure to calculating numbers. And then he went off to Harvard in 1977 to get his MBA. But while he was at Harvard he started working on one of the timesharing programs to help do spreadsheet analysis and wrote his own tool that could do five columns and 20 rows. Then he met Bob Frankston and they added Dan Fylstra, who thought it should be able to run on an Apple - and so they started Software Arts Corporation. Frankston got the programming bug while sitting in on a class during junior high. He then got his undergrad and Masters at MIT, where he spent 9 years in school and working on a number of projects with CSAIL, including Multics. He'd been consulting and working at various companies for awhile in the Boston area, which at the time was probably the major hub. Frankston and Bricklin would build a visible calculator using 16k of space and that could fit on a floppy. They used a time sharing system and because they were paying for time, they worked at nights when time was cheaper, to save money. They founded a company called Software Arts and named their Visual Calculator VisiCalc. Along comes the Apple II. And computers were affordable. They ported the software to the platform and it was an instant success. It grew fast. Competitors sprung up. SuperCalc in 1980, bundled with the Osborne. The IBM PC came in 1981 and the spreadsheet appeared in Fortune for the first time. Then the cover of Inc Magazine in 1982. Publicity is great for sales and inspiring competitors. Lotus 1-2-3 came in 1982 and even Boeing Computer Services got in the game with Boeing Calc in 1985. They extended the ledger metaphor to add sheets to the spreadsheet, which we think of as tabs today. Quattro Pro from Borland copied that feature and despite having their offices effectively destroyed during an earthquake just before release, came to market in 1989. Ironically they got the idea after someone falsely claimed they were making a spreadsheet a few years earlier. And so other companies were building Visible Calculators and adding new features to improve on the spreadsheet concept. Microsoft was one who really didn't make a dent in sales at first. They released an early spreadsheet tool called Multiple in 1982. But Lotus 1-2-3 was the first killer application for the PC. It was more user friendly and didn't have all the bugs that had come up in VisiCalc as it was ported to run on platform after platform. Lotus was started by Mitch Kapor who brought Jonathan Sachs in to develop the spreadsheet software. Kapor's marketing prowess would effectively obsolete VisiCalc in a number of environments. They made TV commercials so you know they were big time! And they were written natively in the x86 assembly so it was fast. They added the ability to add bar charts, pie charts, and line charts. They added color and printing. One could even spread their sheet across multiple monitors like in a magazine. It was 1- spreadsheets, 2 - charts and graphs and 3 - basic database functions. Heck, one could even change the size of cells and use it as a text editor. Oh, and macros would become a standard in spreadsheets after Lotus. And because VisiCalc had been around so long, Lotus of course was immediately capable of reading a VisiCalc file when released in 1983. As could Microsoft Excel, when it came along in 1985. And even Boeing Calc could read Lotus 1-2-3 files. After all, the concept went back to those mainframe delimited files and to this day we can import and export to tab or comma delimited files. VisiCalc had sold about a million copies but that would cease production the same year Excel was released, although the final release had come in 1983. Lotus had eaten their shorts in the market, and Borland had watched. Microsoft was about to eat both of theirs. Why? Visi was about to build a windowing system called Visi-On. And Steve Jobs needed a different vendor to turn to. He looked to Lotus who built a tool called Jazz that was too basic. But Microsoft had gone public in 1985 and raised plenty of money, some of which they used to complete Excel for the Mac that year. Their final release in 1983 began to fade away And so Excel began on the Mac and that first version was the first graphical spreadsheet. The other developers didn't think that a GUI was gonna' be much of a thing. Maybe graphical interfaces were a novelty! Version two was released for the PC in 1987 along with Windows 2.0. Sales were slow at first. But then came Windows 3. Add Microsoft Word to form Microsoft Office and by the time Windows 95 was released Microsoft became the de facto market leader in documents and spreadsheets. That's the same year IBM bought Lotus and they continued to sell the product until 2013, with sales steadily declining. And so without a lot of competition for Microsoft Excel, spreadsheets kinda' sat for a hot minute. Computers became ubiquitous. Microsoft released new versions for Mac and Windows but they went into that infamous lost decade until… competition. And there were always competitors, but real competition with something new to add to the mix. Google bought a company called 2Web Technologies in 2006, who made a web-based spreadsheet called XL2WEB. That would become Google Sheets. Google bought DocVerse in 2010 and we could suddenly have multiple people editing a sheet concurrently - and the files were compatible with Excel. By 2015 there were a couple million users of Google Workspace, growing to over 5 million in 2019 and another million in 2020. In the years since, Microsoft released Office 365, starting to move many of their offerings onto the web. That involved 60 million people in 2015 and has since grown to over 250 million. The statistics can be funny here, because it's hard to nail down how many free vs paid Google and Microsoft users there are. Statista lists Google as having a nearly 60% market share but Microsoft is clearly making more from their products. And there are smaller competitors all over the place taking on lots of niche areas. There are a few interesting tidbits here. One is that the tools that there's a clean line of evolution in features. Each new tool worked better, added features, and they all worked with previous file formats to ease the transition into their product. Another is how much we've all matured in our understanding of data structures. I mean we have rows and columns. And sometimes multiple sheets - kinda' like multiple tables in a database. Our financial modeling and even scientific modeling has grown in acumen by leaps and bounds. Many still used those electro-mechanical calculators in the 70s when you could buy calculator kits and build your own calculator. Those personal computers that flowed out in the next few years gave every business the chance to first track basic inventory and calculate simple information, like how much we might expect in revenue from inventory in stock to now thousands of pre-built formulas that are supported across most spreadsheet tooling. Despite expensive tools and apps to do specific business functions, the spreadsheet is still one of the most enduring and useful tools we have. Even for programmers, where we're often just getting our data in a format we can dump into other tools! So think about this. What tools out there have common file types where new tools can sit on top of them? Which of those haven't been innovated on in a hot minute? And of course, what is that next bold evolution? Is it moving the spreadsheet from a book to a batch process? Or from a batch process to real-time? Or from real-time to relational with new tabs? Or to add a GUI? Or adding online collaboration? Or like some big data companies using machine learning to analyze the large data sets and look for patterns automatically? Not only does the spreadsheet help us do the maths - it also helps us map the technological determinism we see repeated through nearly every single tool for any vertical or horizontal market. Those stuck need disruptive competitors if only to push them off the laurels they've been resting on.
CSAIL's Tim Kraska is developing new autoML approaches to make analytics more accessible to a broader range of users and optimize database architecture systems for companies. Access the transcript for the podcast at: http://cap.csail.mit.edu/sites/default/files/2021-08/Tim%20Kraska%20podcast.pdf.
Intelligenza artificiale, sensori tattili e algoritmi di apprendimento. A cosa servono? A creare il tappeto intelligente per usi sanitari, sportivi e di gioco. Technomondo, la rubrica dedicata alla tecnologia di Raffaella Quadri.
Intelligenza artificiale, sensori tattili e algoritmi di apprendimento. A cosa servono? A creare il tappeto intelligente per usi sanitari, sportivi e di gioco. Technomondo, la rubrica dedicata alla tecnologia di Raffaella Quadri.
Here is your fun fact for the day – Napoleon actually broke the Rosetta Stone. Go figure. In a way, it’s a great metaphor. The Rosetta Stone has been an incredible tool for translating multiple languages in the centuries since its discovery, proving itself a valuable aid in helping put back the pieces of many languages that tend to get broken and lost over time. The value though is not merely in being able to translate ancient languages, it’s in all the history that comes with being able to read ancient texts for the first time. Suddenly a whole perspective on historical events opens up, or knowledge of things we could never have known about otherwise is unlocked. Putting an ancient language back together doesn’t just open up words, it opens up literal worlds. Now, the geniuses over at MIT have come up with another tool that we can use to unlock a few more. A new system has been developed by the Computer Science and Artificial Intelligence Laboratory (CSAIL) that can actually decipher lost languages. Best of all, it doesn’t need extensive knowledge of how it compares with already known languages to crack the code. The program can actually figure out on its own how different languages relate to one another. So, how does that wizardry work? One of the chief insights that make CSAIL’s program possible is the recognition of certain patterns. One of these is that languages only develop in certain ways. Spellings can change in some ways, but not others due to how different certain letters sound. Based on this and other insights, it was possible to develop an algorithm that can pick out a variety of correlations. Of course, such a thing has to be tested before it can be trusted. If you don’t test your language detector, you get bad languages. That’s probably how the whole “Aztecs said the end of the world would be in 2012” thing started. One intern with a bad translator program took it from, “And then I decided I could stop chiseling the years now. I’m a few centuries ahead,” to “the earth will stop completely rotating in 2012”. Fortunately, the researchers at MIT were a bit brighter than that. They took their program and tested it against several known languages, correctly pointing out the relationships between them and putting them in the proper language families. They are also looking to supplement their work with historical context to help determine the meaning of completely unfamiliar words, similar to what most people do when they come across a word they don’t know. They look at the entire sentence and try to figure out the meaning from the surrounding context. Led by Professor Regina Barzilay, the CSAIL team has developed an incredibly useful tool to help us understand not just the events of times gone by, but the way people thought back then. By better understanding the languages of the past, we can learn why people did what they did. We could gain valuable insight into cultures long dead to us. That knowledge will in turn help us to better understand our past and how we got to where we are. It gets us more information, information straight from the source, or at least closer to it. If TARTLE likes anything in the world, it’s getting information straight from the source. After all, that’s what we preach day in and day out around here. Getting our information from the source, minimizing false assumptions and bias when it comes to analyzing information. It’s great to see that same spirit at work in one of the world’s premier research centers and to see it being applied to our past. What’s your data worth? www.tartle.co
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
Today we’re joined by Irene Chen, a Ph.D. student at MIT. Irene’s research is focused on developing new machine learning methods specifically for healthcare, through the lens of questions of equity and inclusion. In our conversation, we explore some of the various projects that Irene has worked on, including an early detection program for intimate partner violence. We also discuss how she thinks about the long term implications of predictions in the healthcare domain, how she’s learned to communicate across the interface between the ML researcher and clinician, probabilistic approaches to machine learning for healthcare, and finally, key takeaways for those of you interested in this area of research. The complete show notes for this episode can be found at https://twimlai.com/go/479.
As one of the founding directors of the Free Software Foundation, CSAIL's Prof. Hal Abelson believes that free software empowers everyone to maintain the freedom to run, edit, contribute to, and share software, as well as see everything the software is doing. Learn more about Prof. Abelson at: https://bit.ly/3fPIEEq. Access the transcript for the podcast at: https://cap.csail.mit.edu/sites/default/files/research-pdfs/ResearcherSpotlight_HalAbelson_02_24_2021.pdf
Andy and Dave discuss the recent announcement that the U.S. Department of Defense announces that it will adopt the Defense Innovation Board’s detailed principles for using AI. The European Commission releases its white paper on AI. The University of Buffalo’s AI Institute receives a grant to study gamers’ brains in order to build AI military robots. Microsoft announces Turing-NLG, a 17-billion parameter language model. MIT’s CSAIL demonstrates TextFooler, which makes synonym-like substitutions of words, the results of which can severely degrade the accuracy of NLP classifiers. Researchers from McAfee show simple tricks to fool Tesla’s Mobileye EyeQ3 camera. And Andy and Dave conclude with a discussion with Professor Josh Bongard, from the University of Vermont, on his recent “xenobots” research.
In this episode, Professor Anant Agarwal spoke with host Dave Finch about the next 5, 10, 25 years of online education for engineers. The keynote session was followed by a live Q&A with the audience. About Anant Agarwal Anant Agarwal is the Founder and CEO of edX. Anant taught the first edX course on circuits and electronics from MIT, which drew 155,000 students from 162 countries. He has served as the director of CSAIL, MIT's Computer Science and Artificial Intelligence Laboratory, and is a professor of electrical engineering and computer science at MIT. He is a successful serial entrepreneur, having co-founded several companies including Tilera Corporation, which created the Tile multicore processor, and Virtual Machine Works. Anant won the Maurice Wilkes prize for computer architecture, and MIT's Smullin and Jamieson prizes for teaching. He is also the 2016 recipient of the Harold W. McGraw, Jr. Prize for Higher Education, which recognized his work in advancing the MOOC movement. Additionally, he is a recipient of the Padma Shri award from the President of India and was named the Yidan Prize for Education Development Laureate in 2018. He held a Guinness World Record for the largest microphone array, and is an author of the textbook "Foundations of Analog and Digital Electronic Circuits." Scientific American selected his work on organic computing as one of 10 World-Changing Ideas in 2011, and he was named in Forbes' list of top 15 education innovators in 2012. Anant, a pioneer in computer architecture, is a member of the National Academy of Engineering, a fellow of the American Academy of Arts and Sciences, and a fellow of the ACM.
Vedere oltre l'opera d'arte e scoprire influenze storiche tra Paesi ed epoche storiche. L'intelligenza artificiale (AI) diventa un mezzo per guardare la storia dell'arte umana da nuove prospettive. Come? Scopritelo nella nuova puntata di Technomondo, la rubrica dedicata alla tecnologia di Raffaella Quadri
Vedere oltre l'opera d'arte e scoprire influenze storiche tra Paesi ed epoche storiche. L'intelligenza artificiale (AI) diventa un mezzo per guardare la storia dell'arte umana da nuove prospettive. Come? Scopritelo nella nuova puntata di Technomondo, la rubrica dedicata alla tecnologia di Raffaella Quadri
In COVID-related AI news, Purdue University has built a website that tracks global response to social distancing, by pulling live footage and images from over 30,000 cameras in 100 countries. Simon Fong, Nilanjan Dey, and Jyotismita Chaki have published Artificial Intelligence for Coronavirus Outbreak, which examines AI’s contribution to combating COVID-19. Researchers at Harvard and Boston Children’s Hospital use a “regular” Bayesian model to identify COVID-19 hotspots over 14 days before they occur. In non-COVID AI news, the acting director of the JAIC announces a shift to enabling joint warfighting operations. The DoD Inspector General releases an Audit of Governance and Protection of DoD AI Data and Technology, which reveals a variety of gaps and weaknesses in AI governance across DoD. Detroit Police Chief James Craig reveals that the police department’s experience with facial recognition technology resulted in misidentified people about 96% of the time. Over 1400 mathematicians sign and deliver a letter to the American Mathematical Society, urging researchers to stop working on predictive-policing algorithms. DARPA awards the Meritorious Public Service Medal to Professor Hava Siegelmann for her creation and research in the Lifelong Learning Machines Program. And Horace Barlow, one of the founders of modern visual neuroscience, passed away on 5 July at the age of 98. In research, Udrescu and Tegmark release AI Feynman 2.0, with unsupervised learning of equations of motion by viewing objects in raw and unlabeled video. Researchers at CSAIL, NVidia, and the University of Toronto create the Visual Causal Discovery Network, which learns to recognize underlying dependency structures for simulated fabrics, such as shirts, pants, and towels. In reports, the Montreal AI Ethics Institute publishes its State of AI Ethics. In the video of the week, Max Tegmark discusses the previously mentioned research on equations of motion, and also discusses progress in symbolic regression. And GanBreeder upgrades to ArtBreeder, which can create realistic-looking images from paintings, cartoons, or just about anything. Click here to visit our website and explore the links mentioned in the episode.
It’s a week of huge announcements! But first, in COVID-related AI news, Andy and Dave discuss a review paper in Chaos, Solitons, and Fractals that provides a more international focus on the role of AI and ML in COVID research. CSAIL teams with Ava Robotics to design a robot that maneuver between waypoints and disinfect surfaces of warehouses with UV-C light. C3.ai Digital Transformation Institute awards $5.4M to 26 AI researchers for projects related to COVID-19. In non-COVID news, the Association for Computing Machinery calls for the immediate suspension of facial recognition technologies until more mature and reliable. US lawmakers have introduced a bill that would ban police use of facial recognition, while separate bills seek to increase the AI talent available for the Department of Defense, and work to realign and rewire the JAIC within DoD. Over 2300 researchers sign a petition to Springer Nature to reject a publication from Harrisburg University, which developed facial recognition software to predict whether somebody was going to be a criminal. Meanwhile, researchers from Stanford demonstrate the problem of reproducibility by giving a data set of brain scans to 70 different researcher teams; no two teams chose the same workflow to analyze the data, and the final conclusions showed a sizeable variation. In a similar vein, researchers at Duke University examine the historical record of brain scan research and find poor correlation across experiments. In research, the “best paper” for the Conference on Computer Vision and Pattern Recognition goes to a team from Oxford, who use unsupervised learning methods and symmetry to convert single 2D images into 3D models. Researchers at Uber, the University of Toronto, and MIT use 3D simulated worlds to generate synthetic data for training LiDAR systems on self-driving vehicles. Calum MacKellar makes Cyborg Mind available, a look into the future of cyberneuroethics. And Johns Hopkins prepares for a second seminar on Operationalizing AI in Health. Click here to visit our website and explore the links mentioned in the episode.
Andy and Dave discuss the initial results from King’s College London’s COVID Symptom Tracker, which found fatigue, loss of taste and smell, and cough to be the most common symptoms. MIT’s CSAIL and clinical team at Heritage Assisted Living announce Emerald, a Wi-Fi box that uses machine learning analyzes wireless signals to record (non-invasively) a person’s vital signs. AI Landing has developed a tool that monitors the distance between people and can send an alert when they get too close. And Johns Hopkins University updates its COVID tracker to provide greater levels of detail on information in the US. In non-COVID news, OpenAI releases Microscope, which contains visualizations of the layers and neurons of eight vision systems (such as AlexNet). The JAIC announces its “Responsible AI Champions” for AI Ethics Principles, and also issues a new RFI for new testing and evaluation technologies. In research, Udrescu and Tegmark publish AI Feynman, and improved algorithm that can find symbolic expressions that match data from an unknown function; they apply the method to 100 equations from Feynman’s Lectures on Physics, and it discovers all of them. The report of the week comes from nearly 60 authors across 30 organizations, a publication on Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims. The review paper of the week provides an overview of the State of the Art on Neural Rendering. The book of the week takes a look at the history of DARPA, in Transformative Technologies: Perspectives on DARPA. Stuart Kauffman gives his thoughts on complexity science and prediction, as they related to COVID-19. The ELLIS society holds its second online workshop on COVID on 15 April. Matt Reed creates Zoombot, a personalized chatbot to take your place in Zoom meetings. Ali Aliev creates Avatarify, to make yourself look like somebody else in real-time for your next Zoom meeting. Click here to visit our website and explore the links mentioned in the episode.
This Week in Machine Learning & Artificial Intelligence (AI) Podcast
Today we’re joined by Aleksander Madry, Faculty in the MIT EECS Department, a member of CSAIL and of the Theory of Computation group. Aleksander, whose work is more on the theoretical side of machine learning research, walks us through his paper “Adversarial Examples Are Not Bugs, They Are Features,” which was published previously presented at last year’s NeurIPS conference. In our conversation, we explore the idea of adversarial examples in machine learning systems being features, with results that might be undesirable, but still working as designed. We talk through what we expect these systems to do, vs what they’re actually doing, if we’re able to characterize these patterns, and what makes them compelling, and if the insights from the paper will inform opinions on either side of the deep learning debate. The complete show notes for this can be found at twimlai.com/talk/369.
丽莎老师讲机器人之MIT让机器人通过感知开始触摸世界欢迎收听丽莎老师讲机器人,想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导,找丽莎老师!欢迎添加微信号:153 5359 2068,或搜索微信公众号:我最爱机器人。麻省理工学院的研究人员首次通过仅利用来自其“感测”皮肤的运动和位置数据,使软机械臂能够了解其在3D空间中的配置,这或许是未来机械臂的又一次变革。这个由柔性极高的材料((类似于在生物体内发现的材料))制作而成的软机器人,也受到目前机器人向着更安全方向的思路影响,比传统刚性机器人的更安全,更柔性,更具弹性和适应性,以及具备生物特性的新一代机器人全新替代品。但是难点同样存在!例如对这些可变形机器人进行自主控制就是一项艰巨的任务,因为这些新软体机器人可以在任何给定时刻沿几乎无限个方向移动,这使得研究人员很难通过编程或者示教,训练用于驱动自动化设备的规划和控制模型。原先,实现自主控制的传统方法(控制系统)是使用具有多个运动捕捉相机的大型视觉系统,该系统能为机器人提供有关3D运动和位置的反馈。但是,目前对于实际应用中的这个新软体机器人而言,这是不切实际的。这是一种配套软体传感器系统,该系统覆盖机器人的身体,以提供“本体感觉”,即感知其身体的运动和位置。该反馈会进入一种新颖的深度学习模型,该模型可筛选出噪声并捕获清晰的信号,以计算机器人的3D位置。研究人员在这个看起来类似于橡树干的软机器人手臂上验证了他们的系统,让该机器人手臂可以自动摆动和伸展,并可以预测自己的空间位置。这些研究人员的软传感器是将导电硅胶片切成折纸形状,使它们具有“压阻”特性,这意味着它们在应变时会改变电阻。当传感器响应机械臂的拉伸和压缩而变形时,其电阻将转换为输出电压,然后将其用作与该运动相关的信号。该传感器可以使用人类现成的材料制造,这意味着今后任何实验室都可以开发自己的系统。研究人员正在感测软机器人特性,使其从传感器(而不是视觉系统)获取反馈以进行控制,而不是和原先一样通过视觉系统进行控制。” “例如,想使用这些柔软的机器人树干来自动定向和控制自己,捡拾东西并与世界互动。向更复杂的自动化控制迈出的第一步。”未来的目标之一是帮助制造出可以更加灵巧地处理和操纵环境中物体的人造肢体。“想想人类自己的身体:您可以闭上眼睛,根据皮肤的反馈来了解世界。” “希望为软机器人设计相同的功能。”完全集成的人体传感器是软机器人技术的长期目标。传统的刚性传感器会损害软机器人的自然柔韧性,使其设计和制造复杂化,并可能导致各种机械故障。因此,基于软材料的传感器是一种更合适的替代方案,但是其设计需要专门的材料和程序运行方法,这使得许多机器人实验室难以在软机器人中制造和集成它们。在CSAIL实验室工作期间,寻找传感器材料的灵感时,研究人员和这些新材料建立了有趣的联系。发现这些用于电磁干扰屏蔽的导电材料薄片可以在任何地方成卷购买。” 这些材料具有“压阻”特性,这意味着它们在应变时会改变电阻。如果将它们放在运动物体上的某些位置,它们可以制成有效的软传感器。当传感器响应于躯干的拉伸和压缩而变形时,能其电阻将转换为特定的输出电压,然后将该电压用作与该运动相关的信号。但是这种材料的伸缩性不高,这将限制其在软机器人中的使用。受折纸的启发,对材料进行切割的启发,设计并用激光切割了矩形的导电硅胶片,将其切割成各种图案,例如成排的小孔或类似链节围栏的交叉切片。这使它们更加灵活,可拉伸,而且“看起来漂亮”。研究人员设计的机器人躯干包括三个部分,每个部分带有四个用于移动手臂的流体致动器(总共12个)。他们在每个段上融合了一个传感器,每个传感器覆盖并收集了来自软机器人中一个嵌入式执行器的数据。他们使用了“等离子键合”技术,该技术可以使一种材料的表面通电,使其与另一种材料键合。大约需要几个小时才能成型出数十个传感器,这些传感器可以使用手持式等离子结合设备结合到软机器人上。如假设的那样,在实验中他们将传感器安在了一个行李箱上,传感器确实捕获了行李箱的总体运动。但是他们真的很吵。“从本质上讲,按我们传统的观念来看,它们在许多方面都是非理想的传感器,因为噪音是工业上非常讨厌的事。”但这只是用软导电材料制造传感器的普遍事实。性能更高且更可靠的传感器需要大多数机器人实验室所没有的专用工具制造。”于是,为了仅使用传感器来估算软机器人的配置,研究人员建立了一个深度神经网络,通过筛查噪声以捕获有意义的反馈信号来完成大部分繁重的工作。研究人员开发了一种新模型,以运动学方式描述了软机器人的形状,从而大大减少了处理模型所需的变量数量。在对比实验中,研究人员让软机器人的躯干摆动,并以随机配置将自己延伸大约一个半小时,他们使用传统的运动捕捉系统获取地面真实数据,同时在训练中,该模型也自主分析了来自其传感器的数据以预测配置,并将其预测与同时收集的地面真实数据进行比较。这样,模型“学习”以将信号模式从其传感器映射到实际配置。结果表明,对于某些更稳定的配置,机器人的传感器估计形状与地面真实情况相符。接下来,研究人员旨在探索新的传感器设计以提高灵敏度,并开发新的模型和深度学习方法,以减少每台新的软机器人所需的训练时间和流程。他们还希望完善系统,以更好地捕获机器人的完整动态运动。当前,该软机器人神经网络和传感器皮肤对捕捉细微运动或动态运动不敏感。但是,对于目前基于学习的软机器人控制方法而言,这是重要的第一步“像我们的软机器人一样,人类的生活系统也不一定是完全精确的,因此,与我们人类相比,机器人一开始也不是精确的机器,做一个看起来不那么精准的机器人,我们无疑做得很好。”
丽莎老师讲机器人之MIT让机器人通过感知开始触摸世界欢迎收听丽莎老师讲机器人,想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导,找丽莎老师!欢迎添加微信号:153 5359 2068,或搜索微信公众号:我最爱机器人。麻省理工学院的研究人员首次通过仅利用来自其“感测”皮肤的运动和位置数据,使软机械臂能够了解其在3D空间中的配置,这或许是未来机械臂的又一次变革。这个由柔性极高的材料((类似于在生物体内发现的材料))制作而成的软机器人,也受到目前机器人向着更安全方向的思路影响,比传统刚性机器人的更安全,更柔性,更具弹性和适应性,以及具备生物特性的新一代机器人全新替代品。但是难点同样存在!例如对这些可变形机器人进行自主控制就是一项艰巨的任务,因为这些新软体机器人可以在任何给定时刻沿几乎无限个方向移动,这使得研究人员很难通过编程或者示教,训练用于驱动自动化设备的规划和控制模型。原先,实现自主控制的传统方法(控制系统)是使用具有多个运动捕捉相机的大型视觉系统,该系统能为机器人提供有关3D运动和位置的反馈。但是,目前对于实际应用中的这个新软体机器人而言,这是不切实际的。这是一种配套软体传感器系统,该系统覆盖机器人的身体,以提供“本体感觉”,即感知其身体的运动和位置。该反馈会进入一种新颖的深度学习模型,该模型可筛选出噪声并捕获清晰的信号,以计算机器人的3D位置。研究人员在这个看起来类似于橡树干的软机器人手臂上验证了他们的系统,让该机器人手臂可以自动摆动和伸展,并可以预测自己的空间位置。这些研究人员的软传感器是将导电硅胶片切成折纸形状,使它们具有“压阻”特性,这意味着它们在应变时会改变电阻。当传感器响应机械臂的拉伸和压缩而变形时,其电阻将转换为输出电压,然后将其用作与该运动相关的信号。该传感器可以使用人类现成的材料制造,这意味着今后任何实验室都可以开发自己的系统。研究人员正在感测软机器人特性,使其从传感器(而不是视觉系统)获取反馈以进行控制,而不是和原先一样通过视觉系统进行控制。” “例如,想使用这些柔软的机器人树干来自动定向和控制自己,捡拾东西并与世界互动。向更复杂的自动化控制迈出的第一步。”未来的目标之一是帮助制造出可以更加灵巧地处理和操纵环境中物体的人造肢体。“想想人类自己的身体:您可以闭上眼睛,根据皮肤的反馈来了解世界。” “希望为软机器人设计相同的功能。”完全集成的人体传感器是软机器人技术的长期目标。传统的刚性传感器会损害软机器人的自然柔韧性,使其设计和制造复杂化,并可能导致各种机械故障。因此,基于软材料的传感器是一种更合适的替代方案,但是其设计需要专门的材料和程序运行方法,这使得许多机器人实验室难以在软机器人中制造和集成它们。在CSAIL实验室工作期间,寻找传感器材料的灵感时,研究人员和这些新材料建立了有趣的联系。发现这些用于电磁干扰屏蔽的导电材料薄片可以在任何地方成卷购买。” 这些材料具有“压阻”特性,这意味着它们在应变时会改变电阻。如果将它们放在运动物体上的某些位置,它们可以制成有效的软传感器。当传感器响应于躯干的拉伸和压缩而变形时,能其电阻将转换为特定的输出电压,然后将该电压用作与该运动相关的信号。但是这种材料的伸缩性不高,这将限制其在软机器人中的使用。受折纸的启发,对材料进行切割的启发,设计并用激光切割了矩形的导电硅胶片,将其切割成各种图案,例如成排的小孔或类似链节围栏的交叉切片。这使它们更加灵活,可拉伸,而且“看起来漂亮”。研究人员设计的机器人躯干包括三个部分,每个部分带有四个用于移动手臂的流体致动器(总共12个)。他们在每个段上融合了一个传感器,每个传感器覆盖并收集了来自软机器人中一个嵌入式执行器的数据。他们使用了“等离子键合”技术,该技术可以使一种材料的表面通电,使其与另一种材料键合。大约需要几个小时才能成型出数十个传感器,这些传感器可以使用手持式等离子结合设备结合到软机器人上。如假设的那样,在实验中他们将传感器安在了一个行李箱上,传感器确实捕获了行李箱的总体运动。但是他们真的很吵。“从本质上讲,按我们传统的观念来看,它们在许多方面都是非理想的传感器,因为噪音是工业上非常讨厌的事。”但这只是用软导电材料制造传感器的普遍事实。性能更高且更可靠的传感器需要大多数机器人实验室所没有的专用工具制造。”于是,为了仅使用传感器来估算软机器人的配置,研究人员建立了一个深度神经网络,通过筛查噪声以捕获有意义的反馈信号来完成大部分繁重的工作。研究人员开发了一种新模型,以运动学方式描述了软机器人的形状,从而大大减少了处理模型所需的变量数量。在对比实验中,研究人员让软机器人的躯干摆动,并以随机配置将自己延伸大约一个半小时,他们使用传统的运动捕捉系统获取地面真实数据,同时在训练中,该模型也自主分析了来自其传感器的数据以预测配置,并将其预测与同时收集的地面真实数据进行比较。这样,模型“学习”以将信号模式从其传感器映射到实际配置。结果表明,对于某些更稳定的配置,机器人的传感器估计形状与地面真实情况相符。接下来,研究人员旨在探索新的传感器设计以提高灵敏度,并开发新的模型和深度学习方法,以减少每台新的软机器人所需的训练时间和流程。他们还希望完善系统,以更好地捕获机器人的完整动态运动。当前,该软机器人神经网络和传感器皮肤对捕捉细微运动或动态运动不敏感。但是,对于目前基于学习的软机器人控制方法而言,这是重要的第一步“像我们的软机器人一样,人类的生活系统也不一定是完全精确的,因此,与我们人类相比,机器人一开始也不是精确的机器,做一个看起来不那么精准的机器人,我们无疑做得很好。”
Today on the podcast, Tom Vander Ark is speaking with Anant Agarwal, the founder and CEO of edX. In 2012, Anant founded edX, which was created in partnership between MIT and Harvard to extend open access to the courses taught by the best professors in the world. On top of this, Anant is also a Professor of Electrical Engineering and Computer Science at MIT. He has also served as the Director of CSAIL, MIT’s Computer Science and Artificial Intelligence Laboratory. In total, he has worked at MIT for 32 years. Presently, he is also a Commission member on the Education Commission and a member of the Board of Directors of the Dana-Farber Cancer Institute. In this episode, Anant shares more of his story and speaks about the wonderful work he’s doing to extend access to more learners through edX. He shares the genesis of edX, why they chose to be a non-profit, some of the history behind MOOCs, some of the new and interesting courses on edX, the future roadmap for edX, and more. Key Takeaways: [:10] About today’s episode. [:37] Tom welcomes Anant Agarwal to the podcast. [:44] Anant speaks about his early education. [1:35] Anant shares what led him to the Indian Institute of Technology Madras. [3:07] Anant speaks about the quality of education he felt he received at IIT Madras. [4:45] After IIT, Anant went to Standford to study Electrical Engineering and Computer Science. He speaks a bit about his experience there. [5:31] Anant speaks about his 32-year career at MIT and the various positions he has served. [6:42] Anant shares the quick origin story of the Computer Science and AI Lab (or CSAIL as it is better known). [7:07] Anant shares the genesis of edX. [9:34] What does Anant believe to be the first real Massive Open Online Course (MOOC)? [11:20] What does Anant think are the pros and cons of making edX an open, non-profit platform? [15:00] Nate McClennon speaks about Getting Smart’s new book, The Power of Place. [16:09] Would Anant say that MOOCs are very much alive and well today? [18:41] Is moving towards shorter skill-oriented certificates and away from degrees a big future trend? [21:27] Are they creating more corporate partnerships with edX? And are they seeing more corporate employers sponsoring online learning? [23:57] Does Anant see the tech giants (such as Amazon, Google, etc.) as new competitors in technical education, partners, or both? [25:32] As a non-profit, does it make it easier to partner with tech giants in technical education? [26:42] How do professionals continuously build tech skills, success/soft skills, and job skills through edX? [28:41] Does edX have any high school partners or high school students on their platform? [30:06] Anant highlights some of the new and interesting courses on edX. [31:22] Anant shares what’s on the roadmap for edX. [33:41] Tom thanks Anant for joining the podcast and for his leadership in this space. Mentioned in This Episode: Anant Agarwal edX MIT Harvard University The Education Commission Coursera MOOC MIT OpenCourseWare Khan Academy The Power of Place: Authentic Learning Through Place-Based Education, by Tom Vander Ark, Dr. Emily Liebtag, and Nate McClennon MicroMasters Programs — edX Get Involved: Check out the blog at GettingSmart.com. Find the Getting Smart Podcast on iTunes, leave a review and subscribe. Is There Somebody You’ve Been Wanting to Learn From or a Topic You’d Like Covered? To get in contact: Email Editor@GettingSmart.com and include ‘Podcast’ in the subject line. The Getting Smart team will be sure to add them to their list!
Andy and Dave discuss the recent announcement that the U.S. Department of Defense announces that it will adopt the Defense Innovation Board’s detailed principles for using AI. The European Commission releases its white paper on AI. The University of Buffalo’s AI Institute receives a grant to study gamers’ brains in order to build AI military robots. Microsoft announces Turing-NLG, a 17-billion parameter language model. MIT’s CSAIL demonstrates TextFooler, which makes synonym-like substitutions of words, the results of which can severely degrade the accuracy of NLP classifiers. Researchers from McAfee show simple tricks to fool Tesla’s Mobileye EyeQ3 camera. And Andy and Dave conclude with a discussion with Professor Josh Bongard, from the Unviersity of Vermont, on his recent “xenobots” research.
丽莎老师讲机器人之MIT的无人机器船协奏曲欢迎收听丽莎老师讲机器人,想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导,找丽莎老师!欢迎添加微信号:153 5359 2068,或搜索微信公众号:我最爱机器人。麻省理工学院计算机科学与人工智能实验室(CSAIL)的研究者们,就开发了一种新型锁闭系统(latching system),可以让无人船克服水流干扰,具备“变形”的新能力。这只自主船只 “组队打野”所带来的想象力,就与单枪匹马的无人船截然不同了。首先,舰队形态拥有加乘级的感知能力,各个船体结合在一起,实现1+1>2的数据协同效果。每个roboat船体都配有传感器、推进器、微处理器、GPS 模块、摄像头等硬件,他们结合在一起,令复杂的通信和控制成为可能,能够在河道、水面实现毫米级精度的连接组合。研究者开发了协调器coordinator和工作器worker。一个或多个worker连接到一个协调器,形成一个 “连接容器平台”(connection -vessel platform, CVP)。每个协调器都知道并可以与所有连接的worker进行无线通信。然后CVP通过比较初始形状和新形状之间的几何差异,使用自定义轨迹规划技术来计算到达目标位置的方式与最短轨迹,并决定要不要移动和分拆。效果是明显的,在 MIT 的游泳池和水流稍微汹涌一些的查尔斯河里进行了测试。机器船通常能够在大约 10 秒内成功连接成队,或者是在几次失败后成功。另外,舰队形态在功能上更加高效灵活。这种3D打印出来的机器船虽然大小仅为之前版本的1/4,但通过定制化锁闭装置(latching mechanism)连接,以无碰撞的路径移动,并重新连接到新集合配置中的适当位置。实际上,在阿姆斯特丹,该机器船舰队正计划实现了晚上收垃圾的操作,它们在运河道中到处游走,定位并连接至有垃圾桶的平台,然后把它们拖回垃圾收集设施,以此让运河重新焕发生机,并通过夜晚操作解放人力。同时,由于形态被改变,矩形的机器船组合在一起,还将收获传统无人船没有的能力——搭建临时水上设施。比如桥梁和舞台,来帮助缓解城市繁忙街道上的拥堵。在麻省理工学院的演示池和计算机模拟中,一组组相连的机器船单元,从直线或正方形重新排列成其他形状,比如矩形和“L”形,整个过程只花了几分钟。研究人员相信,他们的轨迹规划算法能构建更大体积的城市建筑。未来,他们会在阿姆斯特丹市中心的尼莫科学博物馆和正在开发的MARITERETRIN区之间,架起一座横跨60米运河的“动态桥梁”。接完乘客之后,如果发现水道上有东西,这些无人船就会停下来或改道。一旦roboat真正投入使用,白天送人送货,晚上还要加班搞垃圾管理与物流,偶尔还能组装成音乐会舞台、食品市场平台和其他结构。恐怕最勤奋的人类劳模也只能望其项背了。这些船只还可以配备环境传感器,监测城市水域,了解城市和人类健康状况。当然,roboat目前还是一个实验性的项目。但我们发现,无人船正在从单打独斗,变成一个在水面协同战斗的灵敏、动态的智能体,从而完成更多的任务。从这种变化中,我们不难找到一些未来机器人协作的灵感。当机器进入协作时代,城市会变成什么样子?在运河中聚集成群的机器人,会发生什么情况呢?之所以有如此感慨,恐怕是因为这种动态解决方案,能够解答长期以来困囿于无人船的几个问题:第一,能否向更多场景进行延伸?传统的无人船只能在较为宽广的水域进行作业,但随着灵活的可以聚集成各种形状的船只舰队出现,无人船可以成为城市基础设施的重要补充,将一些活动从陆地转移到海洋。一方面能够解决路面拥堵的困境,也让无人船技术有了更大的商业想象力。第二,能否突破技术目标的天花板?既往我们对机器智能的预期,就是和L5自动驾驶汽车、波士顿动力机器人等一样高度智慧体。但这类技术解决方案的弊端也在日益显露,比如投入成本过高,训练周期长,算法难度大,现实落地困难等。而roboat这种灵活的解决方案,核心就是“去掉大脑”,机器人不必具备高智能,只要像蚂蚁一样协同工作,也能完成许多复杂任务。这并不是孤例,实际上已经有科学家让纳米机器人形成“蚁群”,对人体血管进行药物输送与清理,被看做是癌症治疗的新希望。目前看来,让更多的机器船自由组合形成综合智能体,这种务中心化的分布式系统,也让无人船技术变得更加真实了一点。第三,能否实现大规模低成本制造?除了在技术上更具现实价值之外,与传统无人船需要风帆等复杂硬件相比,这种体积更小的roboat完全可以通过3D打印进行生产,价格也更加低廉。通过移动和形态转换的叠加,实现更多复杂的功能,显然更容易打破产业端投资的一些顾虑。未来机器人将按照“无中心分布式系统”模式来运行,大量“愚蠢”的个体在分工的情况下完成高难度的行为。从这个角度看,能闯大洋,能游浅水的无人船,或许正在“蚁群智慧”中成为现实。
Richard Stallman has resigned as president and director of the Free Software Foundation, and that's just one of the major shifts this week. Also what makes Manjaro unique? We chat with one of the founders and find out why it's much more than a desktop environment. Special Guests: Alex Kretzschmar, Bernhard Landauer, Brent Gervais, and Neal Gompa.
A.I. is already better than human doctors at diagnosing skin and breast cancer. And as machine learning advances, it's becoming able to decode more complex information, like brain waves and the human genome. A.I. is beginning to revolutionize medicine, and allowing us to see into the future of our bodies...but can we ever know too much about ourselves? What will happen when machine learning lets us open our own black boxes? In this episode: Physician and author Dr. Siddhartha Mukherjee, Google X founder and Kittyhawk CEO Sebastian Thrun, Regina Barzilay of MIT's J-Clinic and CSAIL, Dr. Andy Schwartz of the University of Pittsburgh, Gill Pratt of the Toyota Research Institute. Learn more about your ad-choices at https://news.iheart.com/podcast-advertisers
In early April, 1999, a time capsule was delivered to the famed architect Frank Gehry with instructions to incorporate it into his designs for the building that would eventually host MIT's Computer Science and Artificial Intelligence Lab, or CSAIL. The time capsule was essentially a museum of early computer history, containing 50 items contributed by the likes of Bill Gates and Tim Berners-Lee.
Aleksander Madry, Associate Professor at CSAIL, tackles key algorithmic challenges in today’s computing as part of his work in the Theory of Computation Group at CSAIL. His work is described as re-thinking machine learning from the perspective of security and robustness. Madry discusses the evolution of the human and machine interaction and provides insight on adoption of M/L systems over the next few years.
丽莎老师讲机器人之折纸“食人花”能抓起超100倍自重的物体欢迎收听丽莎老师讲机器人,想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导,找丽莎老师!受折纸工艺的启发,麻省理工学院计算机科学及人工智能实验室(CSAIL)和哈佛大学的一批研究员设计了一种全新的抓取器,它能够抓住并举起各种形状、尺寸和重量的物体。抓取器的目标就是设计一种会帮你打包行李、杂货的机器人。目前的机械臂能抓取的物品种类极其有限——要么不能太重,要么形状有要求,例如方形、圆柱形等。要展示的抓取器能够拿起并放下各种东西,包括酒瓶、花椰菜、葡萄和鸡蛋等。这款神奇产品外观上是伞状的,在视觉上,与科幻机器人手相比,它与橡胶郁金香花朵的形状。夹子本身是一副骨架结构,由折纸这门古老艺术所启发,覆盖在织物中。连接器将夹具连接到臂上,并且还带有真空管,该真空管从夹具中吸出空气,使其围绕物体折叠。机械臂想要抓取柔性材质的物体通常会很困难,这并不是因为研究人员进行的尝试不够,而是因为刚性的抓取钳本就不适用于精准的抓取工作。一旦抽了真空,花瓣紧皱,骨架折叠,便有了强大的力量,可以紧紧咬住猎物。该设备能够抓取易碎的物体而不会破坏它们,同时仍然保持足够强的抓力,可以抓取比自身重120倍的物体。该团队开发的抓取器呈空心的圆锥状,它包含了三个部分,一个由3D打印制作的、16根硅橡胶制的骨架,一个用于夹持的连接器,以及外面的轻质表皮。它并不是把物体夹住,而是“陷入”物体的表面。它的灵感源于一种名叫“魔球”的折纸工艺品,这种“魔球”由一张长方形的纸折成,能不断在球形和圆柱形之间切换。
Are we even sure what the internet is today? One of the original architects of the internet, David Clark, Senior Research Scientist at CSAIL, talks about shaping the future of the internet, the potential challenges and what it could become.
欢迎收听丽莎老师讲机器人,想要孩子参加机器人竞赛、创意编程、创客竞赛的辅导,找丽莎老师!欢迎添加微信号:153 5359 2068,或搜索微信公众号:我最爱机器人。丽莎老师讲机器人之麻省理工研制出可以用人脑直接控制的机器人。麻省理工大学计算机科学与人工智能实验室(CSAIL)近期研制出一款可以由人脑直接控制的机器人“巴克斯特”,这款机器人实现人脑控制的方式是将人脑电波直接接入“巴克斯特”的智能系统当中,从而实现对该机器人的控制。要实现对机器人“巴克斯特”的控制,需要先戴上一顶特制的“脑电波帽”,将自己的脑电波与“巴克斯特”的人工智能系统相连,在实验当中,一名志愿者坐在巴克斯特对面,并戴好了脑电波帽。之后,工作人员在志愿者与机器人巴克斯特之间的桌子上,放了两个分别标记着“染料”和“电线”的盒子。然后志愿者看着机器人巴克斯特,在脑海里试图将相同的材料放到与这些材料相关的盒子里,即颜料放入颜料盒子、电线放入电线盒。与此同时,巴克斯特将按照志愿者的脑电波完,成志愿者想完成的动作——如果机器人巴克斯特犯了错,则证明志愿者脑电波与其连接错误。经过磨合后,机器人巴克斯特顺利并且正确的实现了志愿者的想法,完成了任务。当然,这只是这项实验的第一步,就目前来看,人工智能机器人巴克斯特的成功,为人脑控制的汽车或假肢等应用提供了无限可能。
Dr. Zoya Bylinskii is a recent PhD graduate from CSAIL at MIT under the supervision of Frédo Durand and Aude Oliva. Her work uses AI to decipher human attention and memory. Where are people looking? What images are most memorable? In this episode we discuss algorithms, neural networks, and a threat bigger than sentient AI […]
Globally, there are approximately 3000 motor vehicle deaths per day, 90% of which are due to human error such as distracted driving or impaired driving (drugs, alcohol, sleep deprivation). With statistics like these, it's no wonder that auto manufacturers are in a race against time, and each other, to develop self-driving cars that will meet the challenges of all driving conditions. Teddy Ort, a researcher and graduate student at MIT's Computer Science & Artificial Intelligence Lab (CSAIL) a part of the Distributed Robotics Laboratory provides an interesting look at the future of self-driving vehicles. The MIT researcher discusses his research program's goal of developing the algorithm and artificial intelligence (AI) necessary to enable a car to avoid all motor vehicle accidents. We'll learn why self-driving cars are not available to the public quite yet. While AI may be perfectly successful within a well-established grid such as a city, it may not score as well in rural areas that are not as delineated through mapping technology. Mr. Ort provides an overview of some of the technical issues that must be overcome before self-driving vehicles rule the roads. Though it may seem sensible that these cars simply do the driving when the tech is able, then hand over the control to a human when needed, data suggests this ‘passing of the reigns' is a sticky problem indeed. The MIT researcher gives an overview of the impediments to rural driving for these self-drivers, and how laser-scanning technology will provide the data necessary to read roads in much the same way a human would. And with unmarked roads comprising approximately 60% of all roads in the US, it's easy to see why camera and laser scanning technology will have to rise to the challenges of rural driving. Further, AI based technology will still have a learning curve when it comes to weather and reflective surfaces, for as a human can easily decipher that a car's image seen in a rainy road reflection is not real, AI must learn this skill. Though challenges certainly lie ahead, Teddy Ort informs us that these self-driving cars are on their way to our garages, but exactly when that day will be remains unknown.
Ted Benson handles machine learning and product strategy for Instabase, a platform for solving and automating complex, organization-spanning data problems. Prior to Instabase, he was the founder and CEO of Cloudstitch (Green Visor; Y Combinator; acquired by Instabase). Ted graduated in 2014 with his Ph.D. from MIT's Computer Science and Artificial Intelligence Laboratory (“CSAIL”). During his time in CSAIL, Ted was advised by David Karger, and worked with other such computer science luminaries as Rob Miller and Tim Berners-Lee (one of the creators of the World Wide Web). Ted was also a member of MIT's Natural Language Processing group. Prior to CSAIL, Ted worked in Industry, including a stint at Google. Ted received his undergraduate degree from the University of Virginia. Simon Yoo is a venture capitalist, and is the Founder and Managing Partner of Green Visor Capital. Based in San Francisco, he and his colleagues are on a mission to help re-invent the financial services industry for the 21st century — one startup at a time. The Green Visor team looks to back passionate entrepreneurs, in the US and abroad, that challenge the status quo — to better promote inclusion, transparency and efficiency — through the innovative use of technology. Simon received his MBA from Cornell University and BA from Kenyon College, where he also served as a Trustee for six years. On this episode of The Fintech Investor Podcast Series, Simon Yoo sits down with Ted Benson to Discuss 1) Ted's background and experiences in MIT's CSAIL program and founding CloudStitch 2) Ted's work at Instabase helping companies ingest their data across various stores of information and make sense of it 3) Ted's views on machine learning today and 4) valuing ideas in the free market vs. academic settings. greenvisorcapital.com/
Back atcha' with another MIT is...Podcast! This month we decided to try something new by bringing in multiple guests onto the show. Let us know if you love it or hate it. We also heavily focused on a certain topic this month: what projects are you involved with inside and outside of the classroom? If you are new here -- each month MIT Student Life presents you with a new podcast episode featuring our students on campus. We talk about life at MIT and get to the bottom of what MIT means to each individual. In this episode, Gonzo '20, Avery '20, Moin '20, and Ethan '20, talk about how they met, what clubs and projects they are currently involved in, and demystify the idea that MIT students just work on PSets all day in their rooms. We hope you enjoy the show! If you want to be a part of the podcast, email studentlifesocial@mit.edu or dangonzo@mit.edu and be sure to follow MITStudents on Instagram (instagram.com/mitstudents) and Snapchat to keep up to date with all things MIT student life! Full Podcast Transcription: http://mitsha.re/4XXn30fpBrh EECS: https://www.eecs.mit.edu/ HackMIT: https://hackmit.org/ MIT UROP: http://uaap.mit.edu/research-exploration/urop TechX: http://techx.io/ MakeMIT: https://makemit.org/ CSAIL: https://www.csail.mit.edu/ Code For Good: http://codeforgood.mit.edu/ MIT Machine Intelligence Community: http://machine-intelligence.mit.edu/ StartLabs: http://startup.mit.edu/
In the second of two guest lectures by Prof. Constantinos Daskalakis of CSAIL, Daskalakis talks about examples of PPAD-completeness reductions, as well as other classes related to PPAD.
In this first of two guest lectures by Prof. Constantinos Daskalakis of CSAIL, Daskalakis talks about the class PPAD, illustrating connections to economic game theory, topology, and graph theory.
MIT researchers have developed a system that promises to make it easier for systems to recover from security breaches.
En este episodio pudimos hablar con Tim Kraska, desde Boston en el laboratorio CSAIL de MIT. Pudimos conversar sobre su carrera en datos, sobre la empresa que fundó (Einblick) para facilitar la colaboración alrededor de los datos que se puede dar en organizaciones con plataformas de low-code / no-code, y también pudimos hablar un poco sobre casos y el futuro de este tipo de tecnologías.
Tenemos a Diego May en Boston atendiendo la Open Data Science Conference. La primera entrevista es con Miguel Paredes. Miguel en este momento trabaja en su doctorado para MIT, en el grupo CSAIL - el laboratorio de ciencia de computación e inteligencia artificial de MIT. Ademas contribuye a las actividades del Centro Martin Trust el cual esta formando una nueva generación de emprendedores con un enfoque en datos.