Multi-Objective Population Based Training
This addresses hyperparameter optimization for multi-objective problems in machine learning, offering an incremental improvement over existing methods.
The authors tackled the problem of hyperparameter optimization for multiple conflicting objectives by introducing MO-PBT, a multi-objective version of Population Based Training, and showed that it outperforms random search, single-objective PBT, and the state-of-the-art MO-ASHA on diverse tasks.
Population Based Training (PBT) is an efficient hyperparameter optimization algorithm. PBT is a single-objective algorithm, but many real-world hyperparameter optimization problems involve two or more conflicting objectives. In this work, we therefore introduce a multi-objective version of PBT, MO-PBT. Our experiments on diverse multi-objective hyperparameter optimization problems (Precision/Recall, Accuracy/Fairness, Accuracy/Adversarial Robustness) show that MO-PBT outperforms random search, single-objective PBT, and the state-of-the-art multi-objective hyperparameter optimization algorithm MO-ASHA.