Fast Extra Gradient Methods for Smooth Structured Nonconvex-Nonconcave Minimax Problems

Authors: Sucheol Lee, Donghwan Kim

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical This paper proposes a two-time-scale EG with anchoring, named fast extragradient (FEG), that has a fast O(1/k2) rate on the squared gradient norm for smooth structured nonconvex-nonconcave problems; the corresponding saddle-gradient operator satisfies the negative comonotonicity condition. This paper further develops its backtracking line-search version, named FEG-A, for the case where the problem parameters are not available. The stochastic analysis of FEG is also provided. and Theorem 4.1. For the L-Lipschitz continuous and ρ-comonotone operator F with ρ > 1/2L and for any z Z (F ), the sequence {zk}k 0 generated by FEG satisfies, for all k 1, F zk 2 <= 4 z0 z 2 / (1/L + 2ρ)^2 k^2.
Researcher Affiliation Academia Sucheol Lee Department of Mathematical Sciences KAIST Daejeon, Republic of Korea Donghwan Kim Department of Mathematical Sciences KAIST Daejeon, Republic of Korea
Pseudocode Yes Algorithm 1 Fast extragradient (FEG) method, Algorithm 2 Fast extragradient method with adaptive step size (FEG-A), Algorithm 3 Stochastic fast extragradient (S-FEG) method
Open Source Code No The paper does not provide any explicit statements about the release of source code or links to a code repository for the described methodology.
Open Datasets No The paper uses a 'Toy example' with a simple quadratic function f(x, y) = ρL^2/2 y^2 for illustration, which is a mathematical construct and not a publicly available dataset with concrete access information.
Dataset Splits No The paper primarily presents theoretical analysis and a mathematical toy example, thus it does not provide specific training/validation/test dataset split information.
Hardware Specification No The paper does not provide any specific hardware details for running experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes For the case ρ = 1/3L and L = 1, Figure 2 illustrates that the FEG converges with an accelerated rate...