Active Learning for Multi-class Image Classification

TL;DR

This paper demonstrates that active learning can significantly reduce the number of training examples needed for CNN-based image classification by strategically selecting high-value samples using uncertainty metrics, especially on complex tasks.

Contribution

It introduces a systematic comparison of uncertainty metrics for active learning in multi-class image classification and validates its effectiveness on multiple datasets.

Findings

Active learning reduces training data requirements.

Uncertainty metrics effectively identify valuable training samples.

More significant improvements are observed on complex classification tasks.

Abstract

A principle bottleneck in image classification is the large number of training examples needed to train a classifier. Using active learning, we can reduce the number of training examples to teach a CNN classifier by strategically selecting examples. Assigning values to image examples using different uncertainty metrics allows the model to identify and select high-value examples in a smaller training set size. We demonstrate results for digit recognition and fruit classification on the MNIST and Fruits360 data sets. We formally compare results for four different uncertainty metrics. Finally, we observe active learning is also effective on simpler (binary) classification tasks, but marked improvement from random sampling is more evident on more difficult tasks. We show active learning is a viable algorithm for image classification problems.