Practical Issues in the Complexity of Neural Networks
Abstract
The equipment purchased under this Grant was used to supplement the theoretical work done under AFOSR-87-0400 with experimental results. The primary use of the equipment was to perform experiments to aid in the generation and testing of theoretical hypotheses about neural networks, regarding the magnitude of synaptic weights, convergence of learning algorithms, computation and learning with bounded-precision analog neural networks, the performance of simulated annealing on structured problems, and the management of replicated data bases. Research is still underway to gather more experimental data and provide theoretical justification for the observations. Keywords: Neural networks, Complexity theory, Fault tolerance, Learning.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 29, 1990
- Accession Number
- ADA221420
Entities
People
- Georg Schnitger
- Ian Parberry
- Piotr Berman
Organizations
- Pennsylvania State University