Problem Specific applications for Neural Networks

Abstract

This thesis examines several topics relating to neural networks. First, the investigation of the error output for a multi-layer perceptron is examined to determine if the error calculation can be modified to decrease the training time. The sigmoid function usually used by multi-layer perceptrons is investigated to determine if modifying terms within the sigmoid function will decrease training time. An improvement is found in the performance by as much as an order of magnitude. The subject of adding an additional class to the problem space is examined. Regardless of the data class added or the network size, there is no advantage to using a previously trained multi-layer perceptron as a starting state to be trained additionally to include the new data class. An investigation to determine whether a multi-layer perceptron can be used to reduce noise added to a signal. Results show that the multi-layer perception, when trained with three specific frequencies, reduced noise but would resonate at these frequencies only. The combination of using a Kohonen self-organizing feature map and a multi-layer perceptron to perform isolated word recognition is tested. An 80 percent accuracy is achieved for speaker dependent, isolated word recognition.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1988
Accession Number
ADA203052

Entities

People

  • Mark K. Lutey

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Air Force
  • Algorithms
  • Automated Speech Recognition
  • Computers
  • Data Sets
  • Electrical Engineering
  • Engineering
  • Frequency
  • Generators
  • Identification
  • Neural Networks
  • Neurons
  • Noise Reduction
  • Pattern Recognition
  • Recognition
  • Sine Waves
  • Word Recognition

Fields of Study

  • Computer science

Readers

  • Computational Modeling and Simulation
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Translation
  • AI & ML - Neural Networks
  • Space