From Dual Tracking to Clipping: Provably Faster Distributionally Robust Multi-Objective Optimization
cs.LG, math.OC
Submitted: 2026-05-07
Updated: 2026-10-05
Comments: 47 pages
License: http://creativecommons.org/licenses/by/4.0/
The gist: Multi-objective optimization (MOO) has received growing attention in applications that require learning under multiple criteria.
Terminology
Abstract
Multi-objective optimization (MOO) has received growing attention in applications that require learning under multiple criteria. However, most existing MOO formulations do not explicitly account for distributional shifts in the data. We introduce distributionally robust multi-objective optimization (DR-MOO), which minimizes multiple objectives under their respective worst-case distributions. We propose Pareto-type solution concepts for DR-MOO and develop multi-gradient descent algorithms (MGDA) with provable guarantees. Leveraging a Lagrangian dual reformulation, we first design a double-loop MGDA that uses an inner loop to estimate dual variables and achieves a total sample complexity O(ε-8) for reaching an ε-Pareto-stationary point. To further improve convergence, we combine large-batch sampling with gradient clipping to accommodate generalized smoothness and control bias in stochastic preference updates, eliminating the need for double sampling. This yields a single-loop double-clip MGDA with substantially improved sample complexity O(ε-4). Our theory applies to nonconvex problems without requiring uniformly bounded gradients of the dual objectives. Experiments demonstrate that our methods are competitive with state-of-the-art MGDA baselines.
Sources
- Unifying Distributionally Robust Optimization via Optimal Transport Theory
- Gradient-Based Multi-Objective Deep Learning: Algorithms, Theories, Applications, and Beyond
- Multi-Task Learning with Deep Neural Networks: A Survey
- Explaining and Harnessing Adversarial Examples
- RotoGrad: Gradient Homogenization in Multitask Learning
- Adam: A Method for Stochastic Optimization
- Stochastic Constrained DRO with a Complexity Independent of Sample Size
- Certifying Some Distributional Robustness with Principled Adversarial Training
- Why gradient clipping accelerates training: A theoretical justification for adaptivity
- MGDA Converges under Generalized Smoothness, Provably
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks