ASTra: A Novel Algorithm-Level Approach to Imbalanced Classification

TL;DR

This paper introduces ASTra, a new algorithm-level method combining a novel activation function and a loss function to improve classification of minority classes in highly imbalanced datasets, showing competitive results.

Contribution

The paper presents ASTra, a novel activation and loss function combination specifically designed for imbalanced classification, outperforming complex ensemble methods in certain scenarios.

Findings

Effective in classifying minority examples with very few instances

Comparable or better results than complex ensemble classifiers

Suitable for highly imbalanced datasets with IRs up to 4000

Abstract

We propose a novel output layer activation function, which we name ASTra (Asymmetric Sigmoid Transfer function), which makes the classification of minority examples, in scenarios of high imbalance, more tractable. We combine this with a loss function that helps to effectively target minority misclassification. These two methods can be used together or separately, with their combination recommended for the most severely imbalanced cases. The proposed approach is tested on datasets with IRs from 588.24 to 4000 and very few minority examples (in some datasets, as few as five). Results using neural networks with from two to 12 hidden units are demonstrated to be comparable to, or better than, equivalent results obtained in a recent study that deployed a wide range of complex, hybrid data-level ensemble classifiers.