A Wasserstein Minimax Framework for Mixed Linear Regression
arXiv:2106.07537
Abstract
Multi-modal distributions are commonly used to model clustered data in statistical learning tasks. In this paper, we consider the Mixed Linear Regression (MLR) problem. We propose an optimal transport-based framework for MLR problems, Wasserstein Mixed Linear Regression (WMLR), which minimizes the Wasserstein distance between the learned and target mixture regression models. Through a model-based duality analysis, WMLR reduces the underlying MLR task to a nonconvex-concave minimax optimization problem, which can be provably solved to find a minimax stationary point by the Gradient Descent Ascent (GDA) algorithm. In the special case of mixtures of two linear regression models, we show that WMLR enjoys global convergence and generalization guarantees. We prove that WMLR's sample complexity grows linearly with the dimension of data. Finally, we discuss the application of WMLR to the federated learning task where the training samples are collected by multiple agents in a network. Unlike the Expectation Maximization algorithm, WMLR directly extends to the distributed, federated learning setting. We support our theoretical results through several numerical experiments, which highlight our framework's ability to handle the federated learning setting with mixture models.
arXiv commentTo appear in 38th International Conference on Machine Learning (ICML 2021)
References in corpus (11)
- NIPS 2016 Tutorial: Generative Adversarial Networks
- Three Approaches for Personalization with Applications to Federated Learning
- Generalization and Equilibrium in Generative Adversarial Nets (GANs)
- Alternating Minimization for Mixed Linear Regression
- Robust Federated Learning: The Case of Affine Distribution Shifts
- On the Convergence and Robustness of Training GANs with Regularized Optimal Transport
- Spectral Experts for Estimating Mixtures of Linear Regressions
- A Convex Duality Framework for GANs
- Distributionally Robust Federated Averaging
- EM Converges for a Mixture of Many Linear Regressions
- Sinkhorn EM: An Expectation-Maximization algorithm based on entropic optimal transport