English

Efficient Algorithms for Sum-of-Minimum Optimization

Optimization and Control 2024-06-11 v2

Abstract

In this work, we propose a novel optimization model termed "sum-of-minimum" optimization. This model seeks to minimize the sum or average of NN objective functions over kk parameters, where each objective takes the minimum value of a predefined sub-function with respect to the kk parameters. This universal framework encompasses numerous clustering applications in machine learning and related fields. We develop efficient algorithms for solving sum-of-minimum optimization problems, inspired by a randomized initialization algorithm for the classic kk-means (Arthur & Vassilvitskii, 2007) and Lloyd's algorithm (Lloyd, 1982). We establish a new tight bound for the generalized initialization algorithm and prove a gradient-descent-like convergence rate for generalized Lloyd's algorithm. The efficiency of our algorithms is numerically examined on multiple tasks, including generalized principal component analysis, mixed linear regression, and small-scale neural network training. Our approach compares favorably to previous ones based on simpler-but-less-precise optimization reformulations.

Keywords

Cite

@article{arxiv.2402.07070,
  title  = {Efficient Algorithms for Sum-of-Minimum Optimization},
  author = {Lisang Ding and Ziang Chen and Xinshang Wang and Wotao Yin},
  journal= {arXiv preprint arXiv:2402.07070},
  year   = {2024}
}