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..
Accelerated Distance-adaptive Methods for Hölder Smooth and Convex Optimization
Authors: Yijin Ren, Haifeng Xu, Qi Deng
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate the performance of our proposed method on a diverse set of convex optimization problems. The goal is to assess its efficiency and robustness across different application scenarios. Additional implementation details and extended results (more different problems and large scale experiments) are provided in the appendix. |
| Researcher Affiliation | Academia | Yijin Ren Haifeng Xu School of Information Management and Engineering Shanghai University of Finance and Economics Antai College of Economics and Management Shanghai Jiao Tong University EMAIL, EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Accelerated Gradient Method with Distance Adaption (AGDA) Algorithm 2 AGDA Line Search Free Modification (AGDA LSFM) Algorithm 3 ̖0 Initialization Method Algorithm 4 r Initialization Method Algorithm 5 Two-Stage Line Search |
| Open Source Code | Yes | Our experimental reproduction scripts will been placed in the supplementary material. We are committed to making our code completely open source. |
| Open Datasets | Yes | We set the constraint radius r = 10 and conduct experiments using real-world datasets from LIBSVM3. For the first test, we use the diabetes dataset to examine robustness. To evaluate performance in non-convex optimization, we trained a Res Net18 model on the CIFAR-10 dataset. 3https://www.csie.ntu.edu.tw/cjlin/libsvm/ |
| Dataset Splits | No | The paper mentions several datasets (LIBSVM, diabetes, Boston housing, CIFAR-10) but does not explicitly provide information about how these datasets were split into training, validation, or test sets, nor does it cite standard splits explicitly within the main text. |
| Hardware Specification | Yes | All experiments were conducted on personal computers, utilizing CPUs for computations, with 16GB of RAM. |
| Software Dependencies | No | The paper mentions the use of optimizers like Adam W and Do G, but does not provide specific version numbers for these or any other key software libraries, programming languages, or solvers. |
| Experiment Setup | Yes | For DOG, we set rϵ = 0.01. Both DADA and AGDA are configured with r = 0.01, while for FGM, we set ϵ = 0.01. We set n = 1000, d = 2000 and µ = 0.005 as the parameters of the problem. We set the constraint radius r = 10 and conduct experiments using real-world datasets from LIBSVM3. For Adam W, we set the learning rate to 10 3 . The Do G algorithm was configured with rϵ = 10 3, which is consistent with the primary hyperparameter used in the AGDA implementation. |