Unsupervised Feature Selection by Heuristic Search with Provable Bounds on Suboptimality
Authors: Hiromasa Arai, Crystal Maung, Ke Xu, Haim Schweitzer
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We implemented and tested the three CSSP-WA* algorithms that were described in Section 3. The experiments described here compare our algorithms to the current state-of-the-art on several large publicly available datasets. |
| Researcher Affiliation | Academia | Hiromasa Arai Dept. of Computer Science University of Texas (Dallas) 800 W Campbell Road Richardson, TX 75080 Hiromasa.Arai@utdallas.edu Crystal Maung Dept. of Computer Science University of Texas (Dallas) 800 W Campbell Road Richardson, TX 75080 Crystal.Maung@gmail.com Ke Xu Dept. of Computer Science University of Texas (Dallas) 800 W Campbell Road Richardson, TX 75080 ke.xu5@utdallas.edu Haim Schweitzer Dept. of Computer Science University of Texas (Dallas) 800 W Campbell Road Richardson, TX 75080 hschweitzer@utdallas.edu |
| Pseudocode | Yes | Figure 1: Example of the subsets graph and the generic heuristic search algorithm. The algorithm maintains the fringe list F and the closed nodes list C. Several choices of f (ni) are discussed in the text. 0. Put the root node into F. 1. While F is nonempty and no solution found: 1.1 Pick ni with the smallest f (ni) from F. 1.2 If ni has k columns return it as the solution. 1.3 Otherwise: 1.3.1 Add ni to C. 1.3.2 Examine all children nj of ni. 1.3.2.1 If nj is in C or F do nothing. 1.3.2.2 Otherwise put nj in F. |
| Open Source Code | No | The paper states 'We implemented and tested the three CSSP-WA* algorithms' but does not provide any link to source code or explicitly state that the code is publicly available. |
| Open Datasets | Yes | The datasets are shown in the following table: Name Size availability Madelon 2, 000 500 UCI Isolet5 1, 559 618 UCI CNAE-9 1, 080 857 UCI Tech TC01 163 29, 261 Technion Day1 20, 000 3, 231, 957 UCI |
| Dataset Splits | No | The paper lists datasets used (Madelon, Isolet5, CNAE-9, Tech TC01, Day1) but does not provide specific information on how these datasets were split into training, validation, or test sets for reproduction, nor does it refer to standard splits with citations. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments (e.g., CPU, GPU models, memory, or specific computing environments). |
| Software Dependencies | No | The paper does not provide specific software dependencies (e.g., library names with version numbers like Python 3.8, PyTorch 1.9) needed to replicate the experiment. |
| Experiment Setup | No | The paper states 'The WA* algorithms use ϵ = 0.5 on Madelon, CNAE-9 and Tech TC01, and ϵ = 0.9 on the other datasets' which is a hyperparameter, but it does not provide comprehensive details on the experimental setup such as other specific hyperparameter values, model initialization, optimizer settings, or training schedules. |