An Interactive Visualization Tool for Understanding Active Learning

TL;DR

This paper introduces an interactive visualization tool that helps users understand the training process of active learning models, compare strategies, and evaluate model performance more effectively.

Contribution

The paper presents a novel interactive visualization tool specifically designed for elucidating active learning processes and comparing different strategies.

Findings

The tool enables visualization of prediction changes over training stages.

Preliminary experiments show the tool's potential in active learning evaluation.

The visualization aids in understanding why certain strategies outperform others.

Abstract

Despite recent progress in artificial intelligence and machine learning, many state-of-the-art methods suffer from a lack of explainability and transparency. The ability to interpret the predictions made by machine learning models and accurately evaluate these models is crucially important. In this paper, we present an interactive visualization tool to elucidate the training process of active learning. This tool enables one to select a sample of interesting data points, view how their prediction values change at different querying stages, and thus better understand when and how active learning works. Additionally, users can utilize this tool to compare different active learning strategies simultaneously and inspect why some strategies outperform others in certain contexts. With some preliminary experiments, we demonstrate that our visualization panel has a great potential to be used in various active learning experiments and help users evaluate their models appropriately.