STOCHASTIC SADDLE POINT PROBLEMS WITH DECISION-DEPENDENT DISTRIBUTIONS

被引:6
作者
Wood, Killian [1 ]
Dall'Anese, Emiliano [1 ,2 ]
机构
[1] Univ Colorado Boulder, Dept Appl Math, Boulder, CO 80309 USA
[2] Univ Colorado Boulder, Dept Elect Comp & Energy Engn, Boulder, CO USA
基金
美国国家科学基金会;
关键词
saddle point problems; minimax problems; stochastic gradient; decision-dependent distributions; OPTIMIZATION; ALGORITHMS; COMMUNICATION; GRADIENT;
D O I
10.1137/22M1488077
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
This paper focuses on stochastic saddle point problems with decision-dependent distributions. These are problems whose objective is the expected value of a stochastic payoff function and whose data distribution drifts in response to decision variables-a phenomenon represented by a distributional map. A common approach to accommodating distributional shift is to retrain optimal decisions once a new distribution is revealed, or repeated retraining. We introduce the notion of equilibrium points, which are the fixed points of this repeated retraining procedure, and provide sufficient conditions for their existence and uniqueness. To find equilibrium points, we develop deterministic and stochastic primal-dual algorithms and demonstrate their convergence with constant step size in the former and polynomial decay step-size schedule in the latter. By modeling errors emerging from a stochastic gradient estimator as sub-Weibull random variables, we provide error bounds in expectation and in high probability that hold for each iteration. Without additional knowledge of the distributional map, computing saddle points is intractable. Thus we propose a condition on the distributional map-which we call opposing mixture dominance-that ensures that the objective is strongly-convex-strongly-concave. Finally, we demonstrate that derivative-free algorithms with a single function evaluation are capable of approximating saddle points.
引用
收藏
页码:1943 / 1967
页数:25
相关论文
共 50 条
  • [21] A Theoretical Revisit to Linear Convergence for Saddle Point Problems
    Wu, Wendi
    Zhao, Yawei
    Zhu, En
    Liu, Xinwang
    Zhang, Xingxing
    Luo, Lailong
    Wang, Shixiong
    Yin, Jianping
    ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2021, 12 (01)
  • [22] On the iterative algorithm for saddle point problems
    Cui, Mingrong
    APPLIED MATHEMATICS AND COMPUTATION, 2008, 204 (01) : 10 - 13
  • [23] A preconditioner for generalized saddle point problems
    Benzi, M
    Golub, GH
    SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2004, 26 (01) : 20 - 41
  • [24] COMPUTATIONAL TECHNIQUES FOR SADDLE POINT PROBLEMS
    Castillo, Zenaida
    Suarez, Jean
    REVISTA INTERNACIONAL DE METODOS NUMERICOS PARA CALCULO Y DISENO EN INGENIERIA, 2008, 24 (03): : 217 - 226
  • [25] A class of smoothers for saddle point problems
    Zulehner, W
    COMPUTING, 2000, 65 (03) : 227 - 246
  • [26] An alternating preconditioner for saddle point problems
    Peng, Xiao-Fei
    Li, Wen
    JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS, 2010, 234 (12) : 3411 - 3423
  • [27] THE SHSS PRECONDITIONER FOR SADDLE POINT PROBLEMS
    Li, Cuixia
    Wu, Shiliang
    JOURNAL OF APPLIED ANALYSIS AND COMPUTATION, 2023, 13 (06): : 3221 - 3230
  • [28] An interior point method for constrained saddle point problems
    Iusem, Alfredo N.
    Kallio, Markku
    COMPUTATIONAL & APPLIED MATHEMATICS, 2004, 23 (01) : 1 - 31
  • [29] Optimal algorithms for differentially private stochastic monotone variational inequalities and saddle-point problems
    Digvijay Boob
    Cristóbal Guzmán
    Mathematical Programming, 2024, 204 : 255 - 297
  • [30] Sparse Polynomial Approximations for Affine Parametric Saddle Point Problems
    Chen, Peng
    Ghattas, Omar
    VIETNAM JOURNAL OF MATHEMATICS, 2023, 51 (01) : 151 - 175