|Tue 18.7||Leila Amgoud|
|Thu 20.7||Jérôme Lang|
|Tue 25.7||Philippe Schlenker|
|Thu 27.7||Simon Thorpe|
All evening lectures are held in Room A - Despax. They start at 19:00, and last around 60 minutes.
Evening lectures are open to all — no participant badge required
Leila Amgoud, IRIT CNRS
Evaluation of arguments: current methods and applications
Tuesday, July 18, 19:00
Argumentation is an alternative approach for defeasible reasoning. It is based on the idea of justifying plausible conclusions by "strong" arguments. Starting from a knowledge base encoded in a logical language, an argumentation system defines arguments and attacks between them using the consequence operator associated with the language. Finally, it uses a semantics for evaluating the arguments. The plausible conclusions to be drawn from the knowledge base are those supported by "good" arguments.
In this talk, we present two families of such systems: the family using extension semantics and the one using ranking semantics. We discuss the outcomes of both families and compare them. We then compare the argumentation approach with other well-known approaches for defeasible reasoning, namely default logic.
Leila Amgoud is a senior CNRS (Centre National de la Recherche Scientifique) researcher at IRIT lab, Toulouse, France. She earned her PhD in computer science from University of Toulouse in 1999. Her research interests include argumentation-based reasoning, nonmonotonic reasoning, inconsistency management, and modeling interactions between autonomous agents (negotiation, persuasion). She serves regularly as program committee member of multiple Artificial Intelligence conferences, and is in the editorial board of the Argument and Computation Journal. She is ECCAI fellow since 2014.
Jérôme Lang, LAMSADE CNRS
Logic, Information, and Computational Social Choice
Thursday, July 20, 19:00.
Social choice theory studies the aggregation of individual preferences towards a collective choice. Computational social choice emerged in the late 1980s, and mostly uses computational paradigms and techniques to provide a better analysis of social choice mechanisms (especially in the fields of voting and of fair division of resources), and to construct new ones.
Logic and information play multiple roles in this interaction between social choice and computer science, among which: representing preferences and reasoning about them; computing collective decisions with incomplete knowledge of agents' preferences; the role of knowledge in strategic behaviour; using logic for automated theorem proving in social choice. The talk will give an overview of part of these topics, and will try to identify challenges for researchers in logic and information (and language?) interested in computational social choice.
Jérôme Lang is a full-time researcher at CNRS (Centre National de la Recherche Scientifique) and is affiliated with LAMSADE, PSL Research University, Université Paris-Dauphine. His interests include computational social choice (in particular voting, fair division, coalition formation, judgment aggregation) and knowledge representation and reasoning (in particular reasoning about beliefs, knowledge and action, as well as preference representation). He is an advisory committee member of Journal of Artificial Intelligence Research, an associate editor of Autonomous Agents and Multi-Agent Systems and of Social Choice and Welfare, and a member of the editorial board of Synthese. He will be the program chair of ECAI-IJCAI 2018. He is a co-editor of the Handbook of Computational Social Choice (Cambridge University Press).
Philippe Schlenker, Institut Jean-Nicod CNRS and New York University
Semantics: where to?
Tuesday, July 25, 19:00
Scientifically, the rise of formal and experimental semantics has been a success story of contemporary linguistics. But the broader impact of semantics has remained limited, particularly in comparison to that of its (in)famous older relative, semiotics. Is it time to broaden the scope of the field?
Philippe Schlenker is a senior researcher at CNRS (Institut Jean-Nicod) and a Global Distinguished Professor at New York University. He was educated at Ecole Normale Supérieure (Paris), and obtained a Ph.D. in Linguistics from MIT, and a Ph.D. in Philosophy from EHESS (Paris). He has taught at Ecole Normale Supérieure, Paris, at the University of Southern California, at UCLA, and at NYU. Dr. Schlenker's early interests include semantics, pragmatics, the philosophy of language and philosophical logic. He has conducted research on indexicals and indirect discourse, intensional semantics ('A Plea for Monsters', Linguistics & Philosophy 2003), anaphora, presuppositions ('Local Contexts', Semantics & Pragmatics 2009), as well as semantic paradoxes. In recent work, Dr. Schlenker has advocated a program of 'super semantics' that seeks to expand the traditional frontiers of the field. He has investigated the semantics of sign languages, with special attention both to their logical structure and to the rich iconic means that interact with it ('Visible Meaning', to appear in Theoretical Linguistics). In order to have a point of comparison for these iconic phenomena, Dr. Schlenker has also investigated the logic and typology of gestures in spoken language. In collaborative work with primatologists and psycholinguists, he has laid the groundwork for a 'primate semantics' that seeks to apply the general methods of formal linguistics to primate vocalizations ('What do Monkey Calls Mean?', Trends in Cognitive Sciences 2016). And in ongoing research, he has has advocated the development of a detailed semantics for music, albeit one that is very different from linguistic semantics ('Outline of Music Semantics', to appear in Music Perception). Dr. Schlenker is the former Managing Editor of Journal of Semantics, and a member of the editorial boards of several semantics journals. His research has been funded by the Fondation Thiers, the American Council of Learned Societies, the NSF, the European Science Foundation, and the European Research Council (Advanced Grant, 2013-2018).
Simon Thorpe, CerCo CNRS
Finding Repeating Structures – the Secret of Intelligence?
Thursday, July 27, 19:00
The last few years have seen Neural Network Architectures based on Deep Learning become the state of the art for many applications, including Vision, Audio and Speech understanding. While such systems can be incredibly powerful, and allow very challenging tasks to be performed using computers equipped with relatively inexpensive yet powerful Graphics processors, their relevance to understanding human intelligence is limited. Deep learning typically requires hundreds of millions of cycles of training with labelled data using supervised learning algorithms such as error back-propagation. In contrast, with some initial experience, a human child may only need a few examples to learn to categorize cats and dogs (for example). In this lecture, I will argue that there are other features of biological learning that can be used as a source of inspiration for artificial systems. In particular, I will argue that the fact that biological neurons use spike-based coding schemes is critical, and that the temporal patterning of activity across populations of neurons may be a vital part of the neural code. I will also describe new unsupervised learning algorithms based on Spike-Time Dependent Plasticity (STDP) that allow neurons to become selective to essentially any repeating spatio-temporal pattern. I will be suggesting that this ability to find repeating patterns of activity in sensory data may be one of the keys to understanding true intelligence, and open new perspectives for artificial intelligence.
Simon Thorpe is a research director with the CNRS. He is the head of the Brain & Cognition Research Center (CerCo) in Toulouse as well being director of the Toulouse Mind & Brain Institute. He studied PPP (Psychology, Philosophy & Psychology) at Oxford, where he also got his PhD with Edmund Rolls. After a postdoc in Canada, he moved to France where he has been working for the CNRS since 1983. For much of his career he has been using a highly interdisciplinary approach to try understand the phenomenal processing speed of biological vision. More recently, his ERC Advanced Grant called "M4: Memory Mechanisms in Man & Machine" has allowed him to investigate the mechanisms that allow the brain to store visual and auditory memories that can last a lifetime.