top
logo

Main Menu

Links


Home Invited Talk
Invited Talk

 

Agents and Data Mining in Bioinformatics: joining data gathering and automatic annotation with classification and distributed clustering.
Ana Lucia C. Bazzan, Universidade Federal do Rio Grande do Sul  -  Instituto de Informatica

Knowledge-Based Reinforcement Learning
Daniel Kudenko, Department of Computer Science, University of York


Title:

Agents and Data Mining in Bioinformatics: joining data gathering and automatic annotation with classification and distributed clustering.

Abstract:

In this talk I will briefly review some works relating agent technologies applied to bioinformatics such as data gathering and automatic annotation.  I will then describe how classification, clustering, and other data mining methods can be useful in annotation of proteins as well as in other tasks related to genomics and proteomics.  I conclude with a discussion about challenges ahead, which relate mainly to conflict resolution (e.g. conflicting annotations), learning from experts (biologists), and computational challenges related to distributed clustering.

Bio:

Ana Bazzan received her PhD in 1997 from  the University of Karlsruhe, Germany. Her previous degrees are in Engineering from the Politechnic School of the University of São Paulo, Brazil, and an MSc. in Computer Science from the Institute of Informatics at the University of Rio Grande do Sul (UFRGS) in Porto Alegre, Brazil. From 1997 to 1998, she had a postdoc research  associate position in the Multi-Agent Systems Laboratory at the University of  Massachusetts in Amherst, under the supervision of Prof. Victor Lesser. In 1999 she joined the Institute for Informatics at UFRGS as an Adjunct Professor and got tenure 3 years later.  She is affiliated with the research groups on Artificial
Intelligence and Multi-Agent Systems at UFRGS. Her research interests include: Game-Theoretic Paradigms for Coordination of Agents, Multiagent Learning, Swarm Intelligence,  Bioinformatics, and Traffic Simulation and Control.  Other professional activities: member of the IFAAMAS (Int. Foundation for Multiagent Systems) board (2004-2008), associate editor of the journal Advances in Complex Systems, president of the steering committee of the special interest group on Artificial Intelligence of
the Brazilian Computing Society (2004-2006), member of the steering committee of the the special interest group on Bioinformatics of the Brazilian Computing Society, PC-chair of the 17th Braz. Symp. on Artificial Intelligence (2004) and of the Brazilian Symposium on Bioinformatics (BSB 2008).


Title:

Knowledge-Based Reinforcement Learning

Abstract:

Reinforcement learning (RL) is a highly popular machine learning technique, mainly due to its natural fit to the agent paradigm and the resulting wide application potential. Nevertheless, RL still suffers from scalability problems, which have prevented its successful use in many complex real-world domains.

For many real-world tasks, human expert knowledge is available. For example, human experts have developed heuristics that help them in planning and scheduling resources. However, this domain knowledge is often rough and incomplete. While not perfect, the knowledge can be used to guide a reinforcement learning agent and restrict its policy search space.

In my talk I will introduce the research area of Knowledge-based Reinforcement Learning and discuss various ways to incorporate domain knowledge into the RL process to speed up the learning, as well as improve the quality of the result.

Bio:

Dr. Daniel Kudenko (http://www.cs.york.ac.uk/~kudenko) is a lecturer in Computer Science at the University of York, UK. His research areas are machine learning and data mining, agents and multi-agent systems, user modeling, and interactive entertainment. In many of these areas he has collaborated with industrial partners in the military and entertainment sector, and has been involved in projects for the Ministry of Defense and Eidos (a major computer games publisher). Dr. Kudenko has served as the coordinator for the ALAD SIG of the AgentLink II network of excellence.

Dr. Kudenko received his Ph.D. in machine learning in 1998 at Rutgers University, NJ. He has participated in several research projects at the University of York, Rutgers University, AT&T Laboratories, and the German Research Center for AI (DFKI) on various topics in artificial intelligence. Dr. Kudenko's work has been published in more than 60 peer-reviewed papers. He has served on numerous program committees and chaired a number of workshops, as well as co-edited three Springer LNCS volumes on Adaptive Agents and Multi-Agent Systems.
 

bottom