Vector Quantization by Minimizing Kullback-Leibler Divergence

TL;DR

This paper introduces a novel vector quantization method that minimizes Kullback-Leibler Divergence to preserve class label information, improving image classification performance.

Contribution

It presents a new objective function and iterative algorithm for vector quantization that better retains class label information compared to traditional methods.

Findings

Enhanced preservation of class label information in quantization

Improved image classification accuracy using the proposed method

Effective minimization of Kullback-Leibler Divergence in quantization

Abstract

This paper proposes a new method for vector quantization by minimizing the Kullback-Leibler Divergence between the class label distributions over the quantization inputs, which are original vectors, and the output, which is the quantization subsets of the vector set. In this way, the vector quantization output can keep as much information of the class label as possible. An objective function is constructed and we also developed an iterative algorithm to minimize it. The new method is evaluated on bag-of-features based image classification problem.