Discrete-Convex-Analysis-Based Framework for Warm-Starting Algorithms with Predictions
Authors: Shinsaku Sakaue, Taihei Oki
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | As for the practical aspect, experiments in Appendix F show that our DCA-based approach performs comparably to (or slightly better than) the method of [17]. |
| Researcher Affiliation | Academia | Shinsaku Sakaue The University of Tokyo Tokyo, Japan sakaue@mist.i.u-tokyo.ac.jp Taihei Oki The University of Tokyo Tokyo, Japan oki@mist.i.u-tokyo.ac.jp |
| Pseudocode | Yes | Algorithm 1 Steepest descent method for L-convex (L -convex) function minimization |
| Open Source Code | No | 3. If you ran experiments... (a) Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [No] All details are described in Appendix F. |
| Open Datasets | No | The instances are generated as follows. We set n = 100 and m = 1000... Edge weights w are randomly sampled from {1, . . . , 100}. (from Appendix F.1). We randomly generate instances of rank-r matroids on a ground set of size n, where elements have weights drawn from {1, . . . , 100}. (from Appendix F.2). We generate image-like data... (from Appendix F.3). No concrete access information for a public dataset is provided. |
| Dataset Splits | No | The paper mentions training on 1000 instances and evaluating on another 1000 instances, implying a train/test split, but does not explicitly define or mention a separate validation set or specific split percentages for validation. |
| Hardware Specification | No | No specific hardware details (like GPU/CPU models or memory) are provided for the experimental setup. |
| Software Dependencies | No | The paper does not provide specific software or library names with version numbers. |
| Experiment Setup | Yes | We set n = 100 and m = 1000 and connect edges randomly with probability 0.1. (Appendix F.1). We train the prediction ˆp using the online gradient descent on 1000 instances with learning rate = 0.1. (Appendix F.1). We use the problem from [28, Section 3.1] with ϕi(pi) = (pi ci)2 and ψij(pj pi) = λ|pj pi|. We generate image-like data of size 100 100 pixels (n = 10000), where ci is sampled uniformly from {0, . . . , 255}, and set λ = 10. (Appendix F.3). |