Neural computing and complex systems (CoSCo)

A complex system is a collection of simple interacting agents, elements or processes whose collective behavior exhibits interesting large scale phenomena. Such systems can be found in various disciplines, including economics, computer science, mathematical biology and physics. Formal models of such systems include e.g., Bayesian belief networks, artificial neural networks, cellular automata, genetic algorithms and autocatalytic networks.

The Complex Systems Computation (CoSCo) research group studies computational issues related to complex systems focusing on learning, model selection, self-organizing behavior and computational complexity questions. The research areas addressed include neural networks, probabilistic and statistical models, cased-based reasoning (CBR) and evolutionary programming (genetic algorithms). The results achieved include

• techniques for mapping Bayesian belief networks to neural network architectures, thus allowing massively parallel implementations of Bayesian reasoning;

• methods for learning probabilistic models from sample data;

• complexity analyses of various problems related to the Hamming and Hopfield neural network models, as well as results about the convergence of genetic algorithms.

• a prototype for a probabilistic DeciSIon enDorsement Environment, D-SIDE;

• an expert system shell (CAse-Based INferencE Tool, CABINET) for case-based reasoning with a probabilistic similarity metrics for case matching and adaptation;

• fundamental studies on the computational power of different neural network models including synchronous, asynchronous, and continuous-time Hopfield nets; on the effects of noise on discrete-time analog computation; and on the computational power of cellular automata -like models (specifically, so called coupled map lattices).

Basic research work by the group has been supported by grants from the Academy of Finland, the University of Helsinki, and various foundations. More applied work has been performed with support from TEKES and the domestic and foreign industrial partners which include, e.g., Kone, Nokia, ABB, and the AT&T and 3M corporations (USA). Some of the resulting software has been adopted in the industry.

Current members of the CoSCo group are Prof. Pekka Orponen, Dr. Petri Myllymäki, M.Sc. Henry Tirri, M.Sc. Tomi Silander, Petri Kontkanen and Jussi Lahtinen. The group has also hosted several foreign graduate students, and participates in two EC-funded research networks (the NeuroCOLT working group, and the Network of Excellence in Neural Computing, NEURONET).

Publications: [41, 72-86, 141-146, 188, 189, 218-225, 229-231].

Home page: http://www.cs.helsinki.fi/research/cosco/