Neural Modelling with Wavelets and Application in Adaptive/Learning Control
Abstract
`Learning' can be viewed as providing an approximation to a desired mapping within a given tolerance. From an electrical engineering perspective, interest in theories of learning arises owing to the presence of many systems that are unknown or only partially known and are difficult to model. In such situations the mapping needs to be implemented from observations during interactions with the system. To solve this problem, researchers in several disciplines have developed tools that can be graphically interpreted as `networks'. Although these tools initially derived some inspiration from biological observations, approximation theory and statistical/information-theoretic methods have been recognized as essential tools to tackle the enormous complexity inherent in the method. Reflecting this diversity of disciplines, and depending on the application domain, such networks are often known variously as `neural networks', `statistical networks', `connectionist networks' and `biological networks'.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1995
- Accession Number
- ADA452462
Entities
People
- Tharmarajah Kugarajah
Organizations
- University of Maryland