Artificial intelligence is having a profound impact as industries increasingly depend on digital resources to innovate. With the future of AI in mind, the AI Summit at re:Invent showcases the latest in AI research and emerging trends. In 30-minute Lightning Talks, attendees hear from leaders in the research community who share their perspectives on everything from AI-fueled cancer research to quantum computing.
View the session catalog to register for the following for the AI Summit (session AIS301):
Delivering on the Promise of AI Together
We are living in a golden age of artificial intelligence (AI). Machines have already surpassed humans in some specific tasks, including image and speech recognition, thanks to the power of cloud computing, the abundance of data required to train AI systems, and improvements in foundational AI algorithms. While some express fear about the potential for AI systems to increasingly overtake the role of humans, together we should influence how these systems can improve every aspect of our lives. Join Rohit Prasad as he explores the opportunities for AI systems to augment human intelligence in ways that will make it accessible to everyone, increasing the societal good today and into the future.
Rohit Prasad is vice president and head scientist for Amazon Alexa, the voice service that powers Amazon’s family of Echo products, Amazon Fire TV, and third-party offerings. Prasad leads Alexa research and development in artificial intelligence technologies aimed at making interaction with Alexa a magical experience for customers. Prior to Amazon, Prasad was deputy manager and senior director of the speech, language, and multimedia business unit at Raytheon BBN Technologies. In that role, he directed U.S. Government-sponsored research and development initiatives in speech-to-speech translation, psychological health analytics, document image translation and STEM learning. Prasad is a named author on more than 100 scientific articles and holds several patents. He earned his master’s degree in electrical engineering from the Illinois Institute of Technology in Chicago and a bachelor’s degree in electronics and communications engineering from Birla Institute of Technology in India.
The Future of Mixed-Autonomy Traffic
How will self-driving cars change urban mobility patterns? This talk examines scientific contributions in the field of reinforcement learning, presented in the context of enabling mixed-autonomy mobility—the gradual and complex integration of autonomous vehicles into existing traffic systems. We explore the potential impact of a small fraction of autonomous vehicles on low-level traffic flow dynamics, using novel techniques in model-free deep reinforcement learning. We share examples in the context of a new open-source computational platform and state-of-the-art microsimulation tools with deep-reinforcement libraries.
Alexandre Bayen is the Liao-Cho Professor of Engineering at UC Berkeley and Director of the Institute of Transportation Studies (ITS). He has published two books and over 200 publications. He holds a PhD and an MS from Stanford University, and a BS from Ecole Polytechnique, France. Bayen has received numerous awards, including the Ballhaus Award from Stanford University, the CAREER award from the National Science Foundation, and he is a NASA Top 10 Innovators on Water Sustainability. He is also the recipient of the Presidential Early Career Award for Scientists and Engineers (PECASE) from the White House. He is also the recipient of the Okawa Research Grant Award, the Ruberti Prize from the IEEE, and the Huber Prize from the ASCE.
Creating Movie Magic with Computational Simulations
Rapid improvements in computational simulation have driven advances in many industries that rely on modelling natural phenomena. In particular, visual effects engineers use AI algorithms to create stunning imagery for feature films like Star Wars, Harry Potter, and the Marvel Cinematic Universe. In this talk, Ronald Fedkiw will discuss his research to produce a new wave of simulation technology with more realistic facial animation to remove the “uncanny valley,” more realistic and predictive cloth simulation, as well as the simulation of botanical trees.
Ronald Fedkiw is a professor at Stanford University, and currently works at the interface between physical simulation and machine learning at the Stanford Artificial Intelligence Laboratory (SAIL). He has been awarded two Academy Awards, the National Academy of Science Award for Initiatives in Research, a Packard Foundation Fellowship, and a Presidential Early Career Award for Scientists and Engineers. He has published over 120 research papers in computational physics, graphics, and vision, and a book on level set methods. For the past 18 years, he has been a consultant with Industrial Light + Magic, receiving screen credits on movies such as “Terminator 3: Rise of the Machines” and “Star Wars: Episode III.”
Unbiased Learning from Biased User Feedback
Logged user interactions are one of the most ubiquitous forms of data available because they can be recorded from a variety of systems (e.g., search engines, recommender systems, ad placement) at little cost. Naively using this data, however, is prone to failure. A key problem lies in biases that systems inject into the logs by influencing where we will receive feedback (e.g., more clicks at the top of the search ranking). This talk explores how counterfactual inference techniques can make learning algorithms robust against bias. This makes log data accessible to a broad range of learning algorithms, from ranking SVMs to deep networks.
Thorsten Joachims is a professor in the Department of Computer Science and the Department of Information Science at Cornell University. His research interests center on a synthesis of theory and system building in machine learning, with applications in search, recommendation, and language technology. His past research focused on counterfactual and causal inference, support vector machines, text classification, structured output prediction, convex optimization, learning to rank, learning with preferences, and learning from implicit feedback. He is an ACM Fellow, AAAI Fellow, and Humboldt Fellow.
Intelligent Systems for Cancer Genomics
One of the most exciting frontiers in science is building automated systems that use existing biomedical data to understand and ultimately treat human disease. The key difficulty in the case of cancer is that it is a highly heterogeneous disease, making it challenging to uncover which molecular alterations in tumors are important for the disease and to predict how an individual will respond to treatment. This talk presents an overview of integrative computational methods for analyzing cancer genomes that leverage a diverse range of complementary data in order to extract biomedically relevant insights.
Mona Singh is a Professor of Computer Science at the Lewis-Sigler Institute for Integrative Genomics at Princeton University. Her research is in computational molecular biology, as well as its interface with machine learning and algorithms with an interest in predicting specificity in protein interactions and uncovering how molecular interactions and networks vary across context, organisms and individuals.
Singh was elected a Fellow of the International Society for Computational Biology in 2018 for outstanding contributions to the fields of computational biology and bioinformatics.
Making Aerial Micro-Drones a Reality
University of Washington
The concept of insect-scale aerial drones has long been in the realm of science fiction rather than reality, especially since powering such small drones is fundamentally difficult. In this talk, Shyam Gollakota will share his work on robofly, the first honeybee-sized wireless drone to successfully lift off. He will also discuss an alternative biology-based solution that integrates sensing, computing, and communication functions onto live-flying bumblebees. Data generated from this mobile IoT platform can feed AI models that have the potential to generate valuable intelligence for applications ranging from precision irrigation to environmental sensing.
Shyam Gollakota is an Associate Professor of the Department of Computer Science & Engineering at the University of Washington where he leads the Networks and Mobile Systems Lab. He is also the President of Jeeva Wireless Inc. His research covers a range of topics, including computer networks, human-computer interaction, battery-free computing and mobile health. He is the recipient of a 2015 National Science Foundation Career Award and an Alfred P. Sloan Fellowship. He is an alumnus of MIT (Ph.D., 2013, winner of ACM doctoral dissertation award) and IIT Madras (2012).
Designing for a Data-Driven Economy
Carnegie Mellon University
The abundance of data available today has been described as a sea change and its own economy. Data has enabled new products, services, businesses, and economies. How can designers thrive as data-savvy innovators in this new economy? What do designers need to know about data, machine learning, and artificial intelligence? In this talk, Jodi Forlizzi draws from multiple research and development efforts to present relevant findings about how to design in a new data-driven economy.
Jodi Forlizzi is the Geschke Director and a professor of Human-Computer Interaction in the School of Computer Science at Carnegie Mellon University. She is responsible for establishing design research as a legitimate form of research in HCI that is different from, but equally as important as, scientific and human science research. For the past 20 years, Jodi has advocated for design research in all forms, mentoring peers, colleagues, and students in its structure and execution. Jodi is a member of the ACM CHI Academy and has been honored by the Walter Reed Army Medical Center for excellence in HRI design research. Jodi has consulted with Disney and General Motors to create innovative product-service systems.
Pragmatic Quantum Machine Learning Today
University of Toronto
Quantum computing’s theoretical potential to exponentially speed up deep learning stands in sharp contrast to the current reality. Implementations are imperfect, suffering from noise and poor coherence times, and scalability limitations. In this talk, we explore how quantum-enhanced machine learning plays a complementary role to classical techniques, rather than acting as a replacement. We discuss relevant computing paradigms, such as quantum annealing and gate-model quantum computing over discrete or continuous variables that are performed efficiently with hybrid classical-quantum protocols.
Peter Wittek is an Assistant Professor at the University of Toronto working on quantum-enhanced machine learning and applications of high-performance learning algorithms in quantum physics. He is also the academic director of the Quantum Machine Learning program in the Creative Destruction lab, a faculty affiliate in the Vector Institute for Artificial Intelligence, and an affiliate in the Perimeter Institute for Theoretical Physics.