An Entropy-based Learning Algorithm of Bayesian Conditional Trees

TL;DR

This paper introduces an entropy-based learning algorithm for Bayesian conditional trees tailored for handwritten digit recognition, grouping similar digits into classes and constructing optimal trees for each class, improving upon previous methods.

Contribution

The paper proposes a modified algorithm that groups similar digits and builds class-specific Bayesian trees, addressing limitations of earlier entropy-based approaches.

Findings

Effective grouping of similar digits improves recognition accuracy.

Constructing class-specific trees reduces model complexity.

Discussion of advantages and extensions enhances understanding of the method.

Abstract

This article offers a modification of Chow and Liu's learning algorithm in the context of handwritten digit recognition. The modified algorithm directs the user to group digits into several classes consisting of digits that are hard to distinguish and then constructing an optimal conditional tree representation for each class of digits instead of for each single digit as done by Chow and Liu (1968). Advantages and extensions of the new method are discussed. Related works of Wong and Wang (1977) and Wong and Poon (1989) which offer a different entropy-based learning algorithm are shown to rest on inappropriate assumptions.