Signal classification using weighted orthogonal regression method

TL;DR

This paper introduces a novel classification method leveraging eigen components and weighted orthogonal regression, improving robustness and performance especially with limited training data.

Contribution

It proposes a new classifier that exploits intrinsic class structures via eigen components and weighted decision criteria, enhancing classification accuracy with small datasets.

Findings

High performance on artificial data

Achieved top results in international competition

Robust classification with limited training data

Abstract

In this paper, a new classifier based on the intrinsic properties of the data is proposed. Classification is an essential task in data mining-based applications. The classification problem will be challenging when the size of the training set is not sufficient to compare to the dimension of the problem. This paper proposes a new classification method that exploits the intrinsic structure of each class through the corresponding Eigen components. Each component contributes to the learned span of each class by specific weight. The weight is determined by the associated eigenvalue. This approach results in reliable learning robust in the case of facing a classification problem with limited training data. The proposed method involves the obtained Eigenvectors by SVD of data from each class to select the bases for each subspace. Moreover, it considers an efficient weighting for the decision-making criterion to discriminate two classes. In addition to high performance on artificial data, this method has increased the best result of international competition.