Learning Environment for Neural Networks and Transient Acoustics
Abstract
As a result of this Phase I research program, we were able to develop a unique workstation which will enable the user to easily interact with the relevant neural network technologies. The Tutor for Underwater Signature Analysis (TUSA) can predict a user's intentions and will provide assistance in an intelligent manner. Along with its intelligence, the system's other strength is its flexibility. TUSA can adapt or can be modified to suit the user's needs. In Phase I, we designed a framework which incorporates the fundamental elements of an intelligent tutor and we showed how these elements interact productively. Future work will complete TUSA's knowledge about backpropagation and underwater signals. More important, however, will be the research into how multiple technologies can be controlled skillfully by a sonar operator to identify signals faster and more accurately. Combining conventional signal processing methods for detection and classification with advanced neural network paradigms will result in a very useful tool. Iterative improvement of TUSA's design through testing with sonar operators will provide a system which will also be applicable to many scientific and engineering fields. Sonar signal analysis, Intelligent tutor, Neural networks, Acoustic signal recognition, Undersea surveillance, Signals, Naval Operations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 14, 1990
- Accession Number
- ADA219000
Entities
People
- Dan Greenwood
- Fareed Stevenson
- Rod Taber
- Steve Deiss