Neural Estimation of Submodular Functions with Applications to Differentiable Subset Selection

Authors: Abir De, Soumen Chakrabarti

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments on synthetic and real data show that FLEXSUBNET outperforms several baselines.
Researcher Affiliation Academia Abir De Soumen Chakrabarti Indian Institute of Technology Bombay {abir,soumen}@cse.iitb.ac.in
Pseudocode No The paper describes the proposed models and methods in detail using mathematical formulations and descriptive text, but does not include any formal pseudocode or algorithm blocks.
Open Source Code Yes Our code is in https://tinyurl.com/flexsubnet. Code is included as part of supplemental material.
Open Datasets Yes We use the Amazon baby registry dataset [30] which contains 17 product categories. ... We generate |V |=104 samples, where we draw the feature vector zs for each sample s V uniformly at random, i.e., zs Unif[0, 1]d with d=10.
Dataset Splits Yes We sample |V |=10000 (set,value) instances as described above and split them into train, dev and test folds of equal size. ... We split S into equal-sized training (Strain), dev (Sdev) and test (Stest) folds.
Hardware Specification No The paper states, "Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)? [Yes] See Appendix F in supplemental material." However, Appendix F is not provided in the main paper, so specific hardware details cannot be extracted.
Software Dependencies No The paper mentions software like BERT and Gumbel-Sinkhorn network but does not provide specific version numbers for these or any other software dependencies used in the experiments.
Experiment Setup Yes Appendix D provides hyperparameter tuning details for all methods. ... Did you specify all the training details (e.g., data splits, hyperparameters, how they were chosen)? [Yes] See Section 5, Appendix D and F in supplemental material.