"Discrete Event Neural Modeling : Toward Linking Biological and Cognitive Levels" Bernard P. Ziegler (Slides)

Theory of modeling and simulation, based on systems theory, aims at proposing a coherent framework for studying and implementing dynamic computational systems. After a brief overview of the theory, representation of biological neural net complexity will be discussed in the context of computational systems. After presenting the usual way of describing the behaviors of neurons (through integrate and fire dynamics) discrete event Markov modeling building blocks will be introduced to be able to connect and implement yourself both conventional and new models of neurons. This approach supports comparison and simplification of neural nets using homomorphic event-based abstraction. With such neural nets constituting the biological layer. A second layer of learning algorithms will be implemented. These novel algorithms take advantage on both group activity and spiking event-like dynamics of neural nets. Finally, the combinations of such blocks will demonstrate how behaviors and structures of logical/cognitive and biological layers might be related to each other.

An extension of the minicourse took place in workshop style including applications on laptops (under windows) using the MS4Me software that allows modeling and simulating in natural language (minimal knowledge of programming language needed).