Издательство MIT Press, 2007, -525 pp.In the course of some 60 to 65 years, going back to the 1940s, signal processing and neural computation have evolved into two highly pervasive disciplines. In their own individual ways, they have significantly influenced many other disciplines. What is perhaps surprising to see, however, is the fact that the cross-fertilization between signal processing and neural computation is still very much in its infancy. We only need to look at the brain and be amazed by the highly sophisticated kinds of signal processing and elaborate hierarchical levels of neural computation, which are performed side by side and with relative ease.
If there is one important lesson that the brain teaches us, it is summed up here:
There is much that signal processing can learn from neural computation, and vice versa.
It is with this aim in mind that in October 2003 we organized a one-week workshop on Statistical Signal Processing: New Directions in the Twentieth Century, which was held at the Fairmont Lake Louise Hotel, Lake Louise, Alberta. To fulfill that aim, we invited some leading researchers from around the world in the two disciplines, signal processing and neural computation, in order to encourage interaction and cross-fertilization between them. Needless to say, the workshop was highly successful.
One of the most satisfying outcomes of the Lake Louise Workshop is that it has led to the writing of this new book. The book consists of 14 chapters, divided almost equally between signal processing and neural computation. To emphasize, in some sense, the spirit of the above-mentioned lesson, the book is entitled New Directions in Statistical Signal Processing: From Systems to Brain. It is our sincere hope that in some measurable way, the book will prove helpful in realizing the original aim that we set out for the Lake Louise Workshop.Modeling the Mind: From Circuits to Systems
Empirical Statistics and Stochastic Models for Visual Signals
The Machine Cocktail Party Problem
Sensor Adaptive Signal Processing of Biological Nanotubes (Ion Channels) at Macroscopic and Nano Scales
Spin Diffusion: A New Perspective in Magnetic Resonance Imaging
What Makes a Dynamical System Computationally Powerful?
A Variational Principle for Graphical Models
Modeling Large Dynamical Systems with Dynamical Consistent Neural Networks
Diversity in Communication: From Source Coding to Wireless Networks
Designing Patterns for Easy Recognition: Information Transmission with Low-Density Parity-Check Codes
Turbo Processing
Blind Signal Processing Based on Data Geometric Properties
Game-Theoretic Learning
Learning Observable Operator Models via the Efficient Sharpening Algorithm
