Interactive Hyperparameter Optimization in Multi-Objective Problems via Preference Learning

TL;DR

Hyperparameter optimization for multi-objective ML is challenged by the need to select a quality indicator for Pareto fronts. The authors propose an interactive, user-centered HPO pipeline that learns a Pareto-front quality indicator from pairwise comparisons using RankSVM, and then optimizes MO-ML hyperparameters with SMAC using the learned indicator. The approach comprises Preliminary Sampling, Interactive Preference Learning, and Utility-driven HPO to personalize front shapes without requiring users to specify an indicator. Experiments in Green AI settings show that the learned-indicator HPO yields substantially better fronts than wrong-indicator baselines and is competitive with advanced users who know which indicator to pick.

Abstract

Hyperparameter optimization (HPO) is important to leverage the full potential of machine learning (ML). In practice, users are often interested in multi-objective (MO) problems, i.e., optimizing potentially conflicting objectives, like accuracy and energy consumption. To tackle this, the vast majority of MO-ML algorithms return a Pareto front of non-dominated machine learning models to the user. Optimizing the hyperparameters of such algorithms is non-trivial as evaluating a hyperparameter configuration entails evaluating the quality of the resulting Pareto front. In literature, there are known indicators that assess the quality of a Pareto front (e.g., hypervolume, R2) by quantifying different properties (e.g., volume, proximity to a reference point). However, choosing the indicator that leads to the desired Pareto front might be a hard task for a user. In this paper, we propose a human-centered interactive HPO approach tailored towards multi-objective ML leveraging preference learning to extract desiderata from users that guide the optimization. Instead of relying on the user guessing the most suitable indicator for their needs, our approach automatically learns an appropriate indicator. Concretely, we leverage pairwise comparisons of distinct Pareto fronts to learn such an appropriate quality indicator. Then, we optimize the hyperparameters of the underlying MO-ML algorithm towards this learned indicator using a state-of-the-art HPO approach. In an experimental study targeting the environmental impact of ML, we demonstrate that our approach leads to substantially better Pareto fronts compared to optimizing based on a wrong indicator pre-selected by the user, and performs comparable in the case of an advanced user knowing which indicator to pick.

Figures (6)

• Figure 1: Overview of the three phases of our approach: Preliminary Sampling provides the user with different Pareto fronts, Interactive Preference Learning allows the user to express their preferences, finally, Utility-driven HPO guides the optimization to the user desiderata.
• Figure 2: Visualization of the feature representation of a Pareto front based on two loss functions.
• Figure 3: Kendall's Tau of the preference learning models.
• Figure 4: Comparison between indicator-based HPO (i.e., IB, columns) and preference-based HPO (i.e., PB, rows). The preference learning model is trained using 28 pairwise comparisons.
• Figure 5: Working example of the DNN wrapper: the grid search over the number of epochs allows us to have a Pareto front as output.