Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Information Theory, Complexity, and Neural Networks
 

Summary: Information Theory, Complexity, and
Neural Networks
Yaser S. Abu-Mostafa
0VER THE PAST FIVE OR SO YEARS, A NEW WAVE
of research in neural networks has emerged. One of the areas
that has attracted a number of researchers is the mathematical
evaluation of neural networks as information processing sys-
tems. In this article, we discuss some ofthe main results in this
area. Researchers have addressed, in specific terms, the ques-
tions of memory capacity, computing power, and learning ca-
pability of different neural network models. The performance
of a conventional computer is usually measured by its speed
and memory. For neural networks, measuring the computing
performance requires new tools from information theory and
computational complexity.
Neural network models offer an interesting alternative to
performing certain computations. They have been considered,
particularly, for unstructured computations, such as pattern
recognition and artificial intelligence problems, and approxi-
mations to large optimization problems.

  

Source: Abu-Mostafa, Yaser S. - Department of Mechanical Engineering & Computer Science Department, California Institute of Technology

 

Collections: Computer Technologies and Information Sciences