Optimistic Tree Searches for Combinatorial Black-Box Optimization
Authors: Cedric Malherbe, Antoine Grosnit, Rasul Tutunov, Haitham Bou Ammar, Jun Wang
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, a numerical assessment is provided to illustrate the potential of tree searches with respect to state-of-the-art methods over typical benchmarks. |
| Researcher Affiliation | Collaboration | Cedric Malherbe1, Antoine Grosnit1, Rasul Tutunov1, Jun Wang1,2, Haitham Bou-Ammar1,2 1Huawei Noah s Ark Lab, 2University College London |
| Pseudocode | Yes | Algorithm 1 Optimistic Lipschitz Tree Search (OLTS) |
| Open Source Code | Yes | Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] and Moreover, all the details regarding baselines, test problems, code and additional plots can be found in Appendix F. |
| Open Datasets | Yes | We compare the performance of OCTS with six methods commonly used to solve combinatorial black-box problems: Simulated annealing [45] (SA), Random Search [9] (RS), Randomized Local Search [38] (RLS), Genetic Algorithm [43] (GA), Evolutionary Algorithm OCTS EA GA RS RLS GHC SA [43] (EA) and Greedy Hill-Climbing [27] (GHC). ... Low autocorrelation binary sequences (LABS) [17]. ... Concatenated trap [17]. ... Maximum independent set (MIS) [17]. ... Max SAT [39]. ... Ising model [17]. ... Contamination [39]. |
| Dataset Splits | No | The paper does not explicitly provide training/validation/test dataset splits. It evaluates performance based on 'Number of function evaluations' and 'Approximation error' without detailing how datasets were partitioned for training, validation, or testing beyond the overall evaluation budget. |
| Hardware Specification | No | The paper explicitly states '[N/A]' for the question 'Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)?' in its checklist. |
| Software Dependencies | No | The paper mentions tools like 'IOHprofiler' and references the 'Max SAT Competition 2018' but does not provide specific version numbers for any software dependencies or libraries used in the experiments. |
| Experiment Setup | Yes | For each problem, we record the best value maxi=1...t f(xi) observed at each iteration 1 t n over ten runs with various dimensionalities d when possible and an evaluation budget set to n = 100 d2 following [18]. |