An optimal linear separator for the Sonar Signals Classification task

TL;DR

This paper demonstrates that sonar signal classification data is linearly separable and provides explicit separating hyperplanes, offering a benchmark for evaluating learning algorithms.

Contribution

It reveals the linear separability of both training and test sets in sonar classification and supplies explicit weights for separating these sets.

Findings

Training and test sets are linearly separable.

Complete set of patterns is linearly separable.

Provides explicit hyperplanes for comparison.

Abstract

The problem of classifying sonar signals from rocks and mines first studied by Gorman and Sejnowski has become a benchmark against which many learning algorithms have been tested. We show that both the training set and the test set of this benchmark are linearly separable, although with different hyperplanes. Moreover, the complete set of learning and test patterns together, is also linearly separable. We give the weights that separate these sets, which may be used to compare results found by other algorithms.