AI at 50

From Programs to Solvers, Models and Techniques for General Intelligence

4pm, Wednesday, 5 December 2007
Psychology Lecture Theatre F21,
7 George Square, Edinburgh

Over the past 20 years, a significant change has occurred in AI research. Many researchers have moved from the early AI paradigm of writing programs for ill-defined problems to writing solvers for well-defined mathematical models such as Constraint Satisfaction Problems, Strips Planning, SAT, Bayesian Networks and Partially Observable Markov Decision Processes.

Solvers are programs that take a compact description of a particular model instance (a planning problem, a CSP instance, and so on) and automatically compute its solution. Unlike the early AI programs, these are general-purpose in the sense that they are not designed to deal with a particular problem but with a large, in fact, infinite collection of problems.

This presents a crisp computational challenge: how to make these solvers scale-up to large and interesting problems given that all these models are intractable in the worst case. Work in these areas has uncovered techniques that accomplish this by automatically recognizing and exploiting the structure of the problem at hand.

My goal in this talk is to articulate this research agenda, to go over some of ideas that underlie these techniques, and to show the relevance of these models and techniques to those interested in models of general intelligence and human cognition.

Hector Geffner is a Distinguished Visiting Researcher in the School of Informatics, and a recently elected Fellow of AAAI. He was a student of Judea Pearl, and won the ACM Distinguished Dissertation Award in 1990. Hector has worked at IBM (Yorktown Heights, NY, USA) and the Universidad Simon Bolivar (Caracas, Venezuela). He is currently a researcher at the Institucion Catalana de Recerca i Estudis Avançats (ICREA) and a professor at the Departamento de Tecnologia, Universitat Pompeu Fabra, where he heads the Artificial Intelligence Group.

People, Computation, and Intelligence

Eric Horvitz, Adaptive Systems and Interaction Group,
Microsoft Research, Redmond

1600, Wednesday, 7 November 2007
Lecture Theatre 2, Appleton Tower,
11 Crichton Street, Edinburgh

Abstract Technical and infrastructural developments are coming together to provide a nurturing environment for creating, studying, and fielding valuable machine learning and reasoning systems. Numerous efforts have been stimulated by the increasing availability of data for studies in learning and adaptation. The data-rich environment poses interesting new challenges and opportunities, and frames new theoretical and practical work. I will present several illustrative research efforts that highlight challenges and directions with the streaming of machine intelligence into the daily lives of people. I will focus thematically on opportunities for harnessing machine learning and reasoning to better understand and support people, and the critical role of methods for representing and reasoning about human intentions, preferences, and initiative.

Biography Eric is a Principal Researcher and Research Area Manager at Microsoft Research. His interests span core challenges in machine reasoning and learning, search and information retrieval, and human-computer interaction. He has served as Associate Editor of the Journal of the ACM, as Chair of the Association for Uncertainty and Artificial Intelligence (AUAI), and on the DARPA Information Science and Technology Study Group (ISAT). He has been elected Councilor and Fellow of the American Association for Artificial Intelligence (AAAI) and is currently serving as President of the organization. He received his PhD and MD degrees at Stanford University.

Rule-based Modeling of Cellular Signalling

Public Lecture

Speaker: Vincent Danos
Title: Rule-based Modeling of Cellular Signalling
Date: Thursday 20th September
Time: 17:00-18:00
Place: e-Science Institute, 13-15 South College Street, Edinburgh

Modelling is becoming a necessity in studying biological signalling pathways, because the combinatorial complexity of such systems rapidly overwhelms intuitive and qualitative forms of reasoning. Yet, this same combinatorial explosion makes the traditional modelling paradigm based on systems of differential equations impractical. In contrast, agent-based or concurrent languages, such as kappa describe biological interactions in terms of rules, thereby avoiding the combinatorial explosion besetting differential equations. Rules are expressed in an intuitive graphical form that transparently represents biological knowledge. In this way, rules become a natural unit of model building, modification, and discussion. We illustrate this with a sizeable example obtained from refactoring two models of EGF receptor signalling that are based on differential equations. An exciting aspect of the agent-based approach is that it naturally lends itself to the identification and analysis of the causal structures that deeply shape the dynamical, and perhaps even evolutionary, characteristics of complex distributed biological systems. In particular, one can adapt the notions of causality and conflict, familiar from concurrency theory, to kappa, our representation language of choice. Using the EGF receptor model as an example, we show how causality enables the formalization of the colloquial concept of pathway and, perhaps more surprisingly, how conflict can be used to dissect the signalling dynamics to obtain a qualitative handle on the range of system behaviours. By taming the combinatorial explosion, and exposing the causal structures and key kinetic junctures in a model, agent- and rule-based representations hold promise for making modelling more powerful, more perspicuous, and of appeal to a wider audience.

Milner Lecture 2007: Ron Fagin

28th June 2007

The 2007 Milner Lecture will be given at 16.00 on 28th June by Ron Fagin of IBM Almaden.