Class Mean Vectors, Self Monitoring and Self Learning for Neural Classifiers

TL;DR

This paper investigates the use of class mean vectors in neural classifiers for direct weight computation, performance monitoring, and self-training, demonstrating promising results on CIFAR-10.

Contribution

It introduces a novel approach leveraging class mean vectors for neural network weight calculation, monitoring, and self-learning, reducing training complexity.

Findings

Effective weight computation without training

Performance monitoring for unlabeled samples

Successful self-training on CIFAR-10

Abstract

In this paper we explore the role of sample mean in building a neural network for classification. This role is surprisingly extensive and includes: direct computation of weights without training, performance monitoring for samples without known classification, and self-training for unlabeled data. Experimental computation on a CIFAR-10 data set provides promising empirical evidence on the efficacy of a simple and widely applicable approach to some difficult problems.