Mapping Learning Algorithms on Data, a useful step for optimizing performances and their comparison

TL;DR

This paper introduces a new methodology for mapping learning algorithms on data to better understand their performance distribution, aiding in selecting optimal configurations and comparing learners across different contexts.

Contribution

The paper presents a novel performance mapping methodology that offers deeper insights into learner behavior and improves comparison across diverse learning contexts.

Findings

Performance maps reveal detailed behavior of algorithms across parameter spaces.

Methodology enhances selection of optimal learner configurations.

Experimental results demonstrate improved comparison of learners.

Abstract

In the paper, we propose a novel methodology to map learning algorithms on data (performance map) in order to gain more insights in the distribution of their performances across their parameter space. This methodology provides useful information when selecting a learner's best configuration for the data at hand, and it also enhances the comparison of learners across learning contexts. In order to explain the proposed methodology, the study introduces the notions of learning context, performance map, and high performance function. It then applies these concepts to a variety of learning contexts to show how their use can provide more insights in a learner's behavior, and can enhance the comparison of learners across learning contexts. The study is completed by an extensive experimental study describing how the proposed methodology can be applied.