Function Prediction Using Recurrent Neural Networks

Abstract

A fully recurrent neural network was applied to the function prediction problem. The real-time recurrent learning (RTRL) algorithm was modified and tested for use as a viable function predictor. The modification gave the algorithm a variable learning rate and a linear/sigmoidal output selection. Verifying the networks ability to temporally learn both the classic exclusive-OR (XOR) problem and the internal state problem, the network was then used to simulate the frequency response of a second order IIR lowpass Butterworth filter. The recurrent network was then applied to two problems: head position tracking, and voice date reconstruction. The accuracy at which the network predicted the pilot's head position was compared to the best linear statistical prediction algorithm. The application of the network to the reconstruction of voice data showed the recurrent network's ability to learn temporally encoded sequences, and make decisions as to whether or not a speech signal sample was considered a fricative or a voiced portion of speech.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1991
Accession Number
ADA243625

Entities

People

  • Randall L. Lindsey

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Air Force
  • Algorithms
  • Artificial Intelligence
  • C Programming Language
  • Computer Programming
  • Computer Programs
  • Computer Science
  • Computers
  • Engineering
  • Literature Surveys
  • Machine Learning
  • Network Science
  • Neural Networks
  • Recurrent Neural Networks
  • Sequences
  • Software Development
  • Three Dimensional

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Neural Network Machine Learning.
  • Speech Processing/Speech Recognition.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks