A simple technique for improving multi-class classification with neural networks

TL;DR

This paper introduces a simple neural network enhancement technique that improves multi-class classification accuracy by adding an additional classification layer on class scores, demonstrated on a 3D hand gesture task.

Contribution

The paper proposes a novel two-stage neural network approach that leverages class score distributions for better class separation, especially in difficult cases.

Findings

Significant accuracy improvement on 10-class gesture recognition

Enhanced disambiguation of hard-to-separate classes

Effective augmentation with raw input boosts performance

Abstract

We present a novel method to perform multi-class pattern classification with neural networks and test it on a challenging 3D hand gesture recognition problem. Our method consists of a standard one-against-all (OAA) classification, followed by another network layer classifying the resulting class scores, possibly augmented by the original raw input vector. This allows the network to disambiguate hard-to-separate classes as the distribution of class scores carries considerable information as well, and is in fact often used for assessing the confidence of a decision. We show that by this approach we are able to significantly boost our results, overall as well as for particular difficult cases, on the hard 10-class gesture classification task.