Semi-Supervised Learning of Classifiers from a Statistical Perspective: A Brief Review

TL;DR

This paper reviews statistical semi-supervised learning methods, highlighting recent findings that classifiers trained on partially labeled data can outperform those trained on fully labeled data in terms of expected error rate.

Contribution

It provides a concise overview of statistical SSL approaches and emphasizes the novel result that partial labels can lead to better classifier performance.

Findings

Classifiers from partially labeled data can have lower expected error rates.

Recent statistical results support the effectiveness of SSL in limited labeled data scenarios.

The review underscores the importance of statistical perspectives in SSL development.

Abstract

There has been increasing attention to semi-supervised learning (SSL) approaches in machine learning to forming a classifier in situations where the training data for a classifier consists of a limited number of classified observations but a much larger number of unclassified observations. This is because the procurement of classified data can be quite costly due to high acquisition costs and subsequent financial, time, and ethical issues that can arise in attempts to provide the true class labels for the unclassified data that have been acquired. We provide here a review of statistical SSL approaches to this problem, focussing on the recent result that a classifier formed from a partially classified sample can actually have smaller expected error rate than that if the sample were completely classified.