New Techniques for Measuring Single Event Related Brain Potentials.

Abstract

Alternative methods of selecting features of visual evoked potentials for automatic pattern classification are compared. Forward sequential feature selection with linear and quadratic discriminant functions, step-wise linear discriminant analysis and exhaustive enumeration with a linear discriminant function are considered. It is found that exhaustive enumeration provides a moderate improvement over the other procedures. In many cases the optimum set of features selected for a given size set does not contain the same features that were found optimum in a set having a different number of features. A new class of filters has been developed for processing evoked potentials that provides much greater discrimination against the ongoing EEG than other filters that have been employed. The filter incorporates all deterministic and statistical information about the signal and noise into a matrix operator that processes the measured data vector. Typical results are shown. A study of the effects of the ongoing EEG on measurement of the latency of evoked potential components is described. The results of the study indicate that the majority of the latency variations found experimentally are intrinsic to the evoked potentials themselves and do not result from interference by the ongoing EEG.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1982
Accession Number
ADA120474

Entities

People

  • C. D. Mcgillem
  • J. I. Aunon

Organizations

  • Purdue University

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Amplitude
  • Bandwidth
  • Data Science
  • Data Sets
  • Discriminant Analysis
  • Electrical Engineering
  • Electrophysiological Phenomena
  • Engineering
  • Feature Selection
  • Information Science
  • Machine Learning
  • Measurement
  • Noise
  • Pattern Recognition
  • Signal Processing
  • Simulations

Readers

  • Neural Network Machine Learning.
  • Neuroscience
  • Regression Analysis.