Self-Interpretable Model with Transformation Equivariant Interpretation

Authors: Yipei Wang, Xiaoqian Wang

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we conduct experiments on image classification tasks with and without transformations. The experiment results demonstrate the high-quality interpretations and the validity of SITE.
Researcher Affiliation Academia Yipei Wang, Xiaoqian Wang Elmore School of Electrical and Computer Engineering Purdue University West Lafayette, IN, 47907 wang4865@purdue.edu, joywang@purdue.edu
Pseudocode No The paper provides mathematical formulations and descriptive text for its model, but it does not include any structured pseudocode or algorithm blocks.
Open Source Code No We will provide the code and instructions at request.
Open Datasets Yes First, we implement SITE on MNIST dataset. and For CIFAR-10, input x is fed to the feature extractor F1. Also, in the questions for reviewers: The data used in the experiments are public available.
Dataset Splits No The test is performed on the untransformed validation set. and Did you specify all the training details (e.g., data splits, hyperparameters, how they were chosen)? [Yes] See the Appendix. The specific split details are deferred to the Appendix, which is not included in the provided text.
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 the beginning of Section 4. However, the beginning of Section 4 and the provided text do not specify any hardware details.
Software Dependencies No The applications of various post-hoc methods are implemented through the Torch Ray toolkit. This mentions a toolkit but does not provide specific version numbers for it or any other software dependencies.
Experiment Setup No The paper states: Did you specify all the training details (e.g., data splits, hyperparameters, how they were chosen)? [Yes] See the Appendix. However, these specific details are not present in the provided main text.