Iterative method improves few-shot multiobjective multitask optimization
Solving Few-Shot Multiobjective Multitask Optimization via Iterative Sequential Transfer
Machine LearningArtificial IntelligenceNeural and Evolutionary Computing
Summary
Optimizing several tasks at once can be tricky, especially when each task has multiple goals and limited chances to try solutions. The authors of this paper focus on multiobjective multitask optimization, where good solutions form a continuous set rather than a single best choice. They propose a new method called Iterative Sequential Transfer that tackles one task at a time and decides which task to work on next based on how ready it is for learning from others. Their tests show this approach works well even when using very few evaluations.
What this means in practice
- •For machine learning engineers: Optimize multiple related models simultaneously with limited evaluation resources by focusing transfers iteratively on tasks ready for knowledge sharing.
- •For operations research teams: Improve efficiency when solving complex multi-criteria, multitask problems under tight resource constraints by sequencing optimization efforts.
Authors
Tingyang Wei, Haofeng Wu, Ananda Phan Iman, Zhao Wei, Jiao Liu, Yew-Soon Ong
Abstract
Applying knowledge transfer across multiple optimization tasks, multitask optimization (MTO) emerges as a promising approach to solving synergistic optimization tasks simultaneously. However, the development of effective knowledge transfer mechanisms in MTO fundamentally relies on aligning elite solution distributions across tasks. This dependency creates a critical bottleneck in few-shot optimization regimes, as restricted evaluation budgets impede the identification of elite solution distributions required for beneficial transfer. This challenge is exacerbated in multiobjective multitask problems, where each optimizer must approximate a continuous Pareto manifold rather than a single optimal point. This paper introduces Iterative Sequential Transfer (IST) to circumvent this bottleneck. We model MTO as a sequence of sequential transfer optimization problems, concentrating evaluations on a single target per iteration. We propose a likelihood-informed task prioritization mechanism to maximize transfer utility by identifying the task most likely ready for knowledge integration. Empirical results on benchmark and real-world problems verify the effectiveness of the proposed method under tight budgets.