A Perceptron-based Fine Approximation Technique for Linear Separation

TL;DR

This paper introduces a perceptron-based approximation technique for efficiently finding separating hyperplanes in binary classification, especially suited for large or imbalanced datasets, by reducing the problem to a one-class classification.

Contribution

It proposes a novel online learning method that tunes neuron weights minimally during hyperplane search, eliminating the need for data labels and bias terms through data transformation.

Findings

Proven convergence of the method.

More efficient than traditional perceptron on large datasets.

Effective for high-dimensional, imbalanced data.

Abstract

This paper presents a novel online learning method that aims at finding a separator hyperplane between data points labelled as either positive or negative. Since weights and biases of artificial neurons can directly be related to hyperplanes in high-dimensional spaces, the technique is applicable to train perceptron-based binary classifiers in machine learning. In case of large or imbalanced data sets, use of analytical or gradient-based solutions can become prohibitive and impractical, where heuristics and approximation techniques are still applicable. The proposed method is based on the Perceptron algorithm, however, it tunes neuron weights in just the necessary extent during searching the separator hyperplane. Due to an appropriate transformation of the initial data set we need not to consider data labels, neither the bias term. respectively, reducing separability to a one-class classification problem. The presented method has proven converge; empirical results show that it can be more efficient than the Perceptron algorithm, especially, when the size of the data set exceeds data dimensionality.