Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

A Near-Optimal Algorithm for Decentralized Convex-Concave Finite-Sum Minimax Optimization

Authors: Hongxu Chen, Ke Wei, Haishan Ye, Luo Luo

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, numerical experiments are conducted to evaluate the performance of Algorithm 2. Numerical experiments demonstrate that our algorithm outperforms existing methods in practice.
Researcher Affiliation Collaboration 1School of Data Science, Fudan University 2School of Management, Xi an Jiaotong University 3SGIT AI Lab, State Grid Corporation of China 4Shanghai Key Laboratory for Contemporary Applied Mathematics
Pseudocode Yes Algorithm 1 Fast Mix(U0, W, R) Algorithm 2 DIVERSE
Open Source Code Yes The code is included in the supplementary material.
Open Datasets Yes The numerical experiments are conducted on datasets a9a, w8a, ijcnn1, and cod-rna, from the LIBSVM repository [14].
Dataset Splits No The paper mentions distributing samples across nodes but does not specify how the overall datasets (a9a, w8a, ijcnn1, cod-rna) are split into training, validation, or test sets for the experiments.
Hardware Specification No It requires few computational resources to run the numerical experiments.
Software Dependencies No The paper does not provide specific version numbers for software components or libraries used in the experiments.
Experiment Setup Yes The regularization parameters are set to be r1 = r2 = 0.2. The parameters of these algorithms are set according to the theoretical analysis or the recommended settings by the authors [39, 46, 55]. Specifically, the parameter b in the DIVERSE is set to be 128 and the fixed batch size for each node in OADSVI is set to be 3. The best performance step sizes from {0.1, 0.05, 0.01} are used, up to the algorithms and the datasets.