Controllable Pareto Front Learning and Applications

Tóm tắt báo cáo

In the evolving landscape of optimization and machine learning, the concept of Pareto Front Learning (PFL) has emerged as a pivotal tool for understanding and navigating trade-offs between conflicting objectives. This talk will delve into the details of PFL, a technique that identifies a set of optimal solutions where improving one objective cannot be achieved without worsening at least one other. We will explore the recent advancements and methodologies in this area, with a focus on the controllability aspects of learning the Pareto fronts. Furthermore, we will examine the applications of controllable PFL in diverse domains such as multi-task learning, multi-objective recommendations, and engineering design.