Mitigating Gradient Bias in Multi-objective Learning: A Provably Convergent Stochastic Approach
Heshan Fernando, Han Shen, Miao Liu, Subhajit Chaudhury, Keerthiram Murugesan, Tianyi Chen
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/heshandevaka/Trade-Off-MOLpytorch★ 15
- github.com/MindSpore-scientific/code-8/tree/main/PCGrad-mindspore-examplemindspore★ 0
- github.com/MindSpore-scientific/code-4/tree/main/PCGrad-mindspore-examplemindspore★ 0
- github.com/MindSpore-scientific/code-1/tree/main/PCGrad-mindspore-examplemindspore★ 0
Abstract
Machine learning problems with multiple objective functions appear either in learning with multiple criteria where learning has to make a trade-off between multiple performance metrics such as fairness, safety and accuracy; or, in multi-task learning where multiple tasks are optimized jointly, sharing inductive bias between them. This problems are often tackled by the multi-objective optimization framework. However, existing stochastic multi-objective gradient methods and its variants (e.g., MGDA, PCGrad, CAGrad, etc.) all adopt a biased noisy gradient direction, which leads to degraded empirical performance. To this end, we develop a stochastic Multi-objective gradient Correction (MoCo) method for multi-objective optimization. The unique feature of our method is that it can guarantee convergence without increasing the batch size even in the non-convex setting. Simulations on multi-task supervised and reinforcement learning demonstrate the effectiveness of our method relative to state-of-the-art methods.