A Psychological Theory of Explainability

Authors: Scott Cheng-Hsin Yang, Nils Erik Tomas Folke, Patrick Shafto

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

Reproducibility Variable Result LLM Response
Research Type Experimental A pre-registered user study on AI image classifications with saliency map explanations demonstrate that our theory quantitatively matches participants predictions of the AI.
Researcher Affiliation Academia 1Department of Mathematics and Computer Science, Rutgers University Newark, New Jersey, USA 2School of Mathematics, Institute for Advanced Study, New Jersey, USA.
Pseudocode No The paper describes mathematical formulations and processes but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes All experiments, mathematical models, analysis code, and hypothesis tests were preregistered (https://osf.io/ 4n67p).
Open Datasets Yes Image Net (Russakovsky et al., 2015), misclassified images drawn from Image Net, and misclassified images drawn from the Natural Adversarial Image Net dataset (Hendrycks et al., 2021).
Dataset Splits Yes To compare the predictive performance of the full model to the alternatives, we used leave-one-out cross-validation (LOO-CV) to control for model complexity.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments, such as GPU models, CPU types, or memory specifications.
Software Dependencies No The paper mentions the use of a 'Res Net-50 model' but does not provide any specific software versions for libraries, frameworks, or other dependencies used in the experiments.
Experiment Setup No The paper describes the mathematical formulations of its models and the method used to fit a parameter (λ), but it does not provide specific hyperparameter values or detailed system-level training settings.