Multi-Task Learning as a Bargaining Game
multi-task-learning
deeplearning
pytorch
In this article, we explore gradient combination in multi-task learning (MTL) as a cooperative bargaining game, and discuss Nash MTL — a novel approach — in detail.