Title: Grandmaster level in StarCraft II using multi-agent reinforcement learning.
Abstract: Many real-world applications require artificial agents to compete and coordinate with other agents in complex environments. As a stepping stone to this goal, the domain of StarCraft has emerged as an important challenge for artificial intelligence research, owing to its iconic and enduring status among the most difficult professional esports and its relevance to the real world in terms of its raw complexity and multi-agent challenges. Over the course of a decade and numerous competitions the strongest agents have simplified important aspects of the game, utilized superhuman capabilities, or employed hand-crafted sub-systems. Despite these advantages, no previous agent has come close to matching the overall skill of top StarCraft players. We chose to address the challenge of StarCraft using general-purpose learning methods that are in principle applicable to other complex domains: a multi-agent reinforcement learning algorithm that uses data from both human and agent games within a diverse league of continually adapting strategies and counter-strategies, each represented by deep neural networks. We evaluated our agent, AlphaStar, in the full game of StarCraft II, through a series of online games against human players. AlphaStar was rated at Grandmaster level for all three StarCraft races and above 99.8% of officially ranked human players.
Short bio: Oriol Vinyals is a Principal Scientist at Google DeepMind, working in deep learning and artificial intelligence. Prior to joining DeepMind, Oriol was part of the Google Brain team. He holds a Ph.D. in EECS from the University of California, Berkeley and is a recipient of the 2016 MIT TR35 innovator award. His research has been featured multiple times at the New York Times, Financial Times, WIRED, BBC, etc., and his articles have been cited over 50,000 times. Some of his contributions are used in Google Translate, Text-To-Speech, and Speech recognition, serving billions of queries every day, and he was the lead researcher of the AlphaStar project, creating an agent that defeated a top professional at the game of StarCraft, achieving Grandmaster level.
Title: Emotion recognition
Abstract: The interest of technologies for emotion recognition is growing rapidly because of their applications in areas like health, user experience, entertainment, or human assistance. However, how can machines recognize emotions? In this talk I will give a brief introduction to Emotional AI and will present some ongoing research in this area. In particular, I will talk about emotion recognition systems based on Computer Vision. Then, I will talk about conversational agents and the importance of incorporating emotional awareness to improve the quality of dialog systems. I will also present a new evaluation methodology to better approximate human evaluations of conversational agents.
Short bio: Àgata Lapedriza is a Professor at the Universitat Oberta de Catalunya. She received her MS degree in Mathematics at the Universitat de Barcelona and her Ph.D. degree in computer science at the Computer Vision Center, at the Universitat Autònoma Barcelona. She was working as a Visiting Researcher in the Computer Science and Artificial Intelligence Lab, at the Massachusetts Institute of Technology (MIT), from 2012 until 2015. Currently she is also a Visiting Researcher at the Affective Computing Group at MIT Medialab, where she leads the project of Emotion Recognition in Context. Her research interests are related to computer vision (image and scene understanding), natural language processing, emotional artificial intelligence, explainable AI, and fairness in AI. She is also interested in the applications of these topics to emotional wellbeing, social robotics, and education.
Title: How does deep learning work?
Abstract: Machine learning algorithms identify hopefully-appropriate functional mappings between input and output variables, f(x)=y. Given a large, well-curated dataset of corresponding x and y pairs, one can estimate the performance of the learned functional mapping f(). But, at some point, one must release the algorithm into “the wild.” The main question in any developer’s mind before release is “how will my algorithm fare under completely unconstrained, unknown conditions?” This talk will present a new set of tools my group has developed to answer this question. Our approach can be used to train a deep learning algorithm without worrying about overfitting, give an estimate on how the algorithm will perform under unknown conditions, and even detect adversarial attacks.
Short bio: Aleix M. Martinez is a Professor in the Department of Electrical and Computer Engineering at The Ohio State University (OSU). Prior to joining OSU, he was with the Department of Electrical and Computer Engineering at Purdue University, and a Research Scientist at the Sony Computer Science Lab. Aleix is most known for being the first to define many problems and solutions in face recognition, discriminant analysis, structure from motion, demonstrating the existence of a much larger set of cross-cultural facial expressions of emotion than previously known (i.e., compound emotion) as well as the transmission of emotion through changes in facial color, and defining a new algebraic topology approach to explaining how deep networks learn to generalize. He has received best paper awards at CVPR and ECCV, a Google Faculty Research award, a Lumely Research Award, and, from 2012-2018, he served as a member of NIH’s Cognition and Perception study section.
Title: Gender bias and natural language processing.
Abstract: Demographic biases are widely affecting artificial intelligence. In particular, gender bias is clearly spread in natural language processing applications, e.g. from stereotyped translations to poorer speech recognition for women than for men. In this talk, I am going to overview the research and challenges that are currently emerging towards fairer natural language processing in terms of gender.
Short bio: Marta R. Costa-jussà is a Ramon y Cajal Researcher at Universitat Politècnica de Catalunya (UPC). She received her PhD from the UPC in 2008. Her research experience is mainly in machine translation. She has worked at LIMSI-CNRS (Paris), Barcelona Media Innovation Center, Universidade de São Paulo, Institute for Infocomm Research (Singapore), Instituto Politécnico Nacional (Mexico), and at University of Edinburgh. Recently, she has received the Google Faculty Research Award.