Understanding Instance-based Interpretability of Variational Auto-Encoders

Authors: Zhifeng Kong, Kamalika Chaudhuri

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we evaluate VAE-Trac In on several real world datasets with extensive quantitative and qualitative analysis.5 Experiments In this section, we aim to answer the following questions. Does VAE-Trac In pass a sanity check for instance-based interpretations? Which training samples have the highest and lowest self influences, respectively? Which training samples have the highest influences over (i.e. are strong proponents of) a test sample? Which have the lowest influences over it (i.e. are its strong opponents)? These questions are examined in experiments on the MNIST [Le Cun et al., 2010] and CIFAR-10 [Krizhevsky et al., 2009] datasets.
Researcher Affiliation Academia Zhifeng Kong Computer Science and Engineering University of California San Diego La Jolla, CA 92093 z4kong@eng.ucsd.edu Kamalika Chaudhuri Computer Science and Engineering University of California San Diego La Jolla, CA 92093 kamalika@cs.ucsd.edu
Pseudocode No The paper provides mathematical derivations and equations for VAE-Trac In but does not present a structured pseudocode block or algorithm.
Open Source Code Yes 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)? [Yes] We will include code in the supplemental material.
Open Datasets Yes These questions are examined in experiments on the MNIST [Le Cun et al., 2010] and CIFAR-10 [Krizhevsky et al., 2009] datasets.
Dataset Splits Yes 3. If you ran experiments... (b) Did you specify all the training details (e.g., data splits, hyperparameters, how they were chosen)? [Yes] See Appendix B.2.
Hardware Specification Yes 3. If you ran experiments... (d) 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] We use one NVIDIA 2080Ti GPU.
Software Dependencies No The paper mentions training details are in Appendix B.2, but does not explicitly list specific software dependencies with version numbers in the main text or the checklist provided.
Experiment Setup Yes 3. If you ran experiments... (b) Did you specify all the training details (e.g., data splits, hyperparameters, how they were chosen)? [Yes] See Appendix B.2.