LGAICVAug 4, 2022

HPO: We won't get fooled again

arXiv:2208.03320v1h-index: 31
Originality Incremental advance
AI Analysis

This work addresses a potential bias in HPO pipelines for machine learning practitioners, though it is incremental as it builds on existing fitness landscape analysis methods.

The study investigated whether the hyperparameter optimization (HPO) pipeline biases the HPO landscape, finding that in most instances, diverse hyperparameters yield the same poor performance, likely due to majority class prediction models, and this worsens the correlation between observed and neighborhood fitness, hindering local-search strategies.

Hyperparameter optimization (HPO) is a well-studied research field. However, the effects and interactions of the components in an HPO pipeline are not yet well investigated. Then, we ask ourselves: can the landscape of HPO be biased by the pipeline used to evaluate individual configurations? To address this question, we proposed to analyze the effect of the HPO pipeline on HPO problems using fitness landscape analysis. Particularly, we studied the DS-2019 HPO benchmark data set, looking for patterns that could indicate evaluation pipeline malfunction, and relate them to HPO performance. Our main findings are: (i) In most instances, large groups of diverse hyperparameters (i.e., multiple configurations) yield the same ill performance, most likely associated with majority class prediction models; (ii) in these cases, a worsened correlation between the observed fitness and average fitness in the neighborhood is observed, potentially making harder the deployment of local-search based HPO strategies. Finally, we concluded that the HPO pipeline definition might negatively affect the HPO landscape.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes