JuliaTrustworthyAI/CounterfactualTraining.jl

Code and research results for our IEEE SaTML 2026 paper "Counterfactual Training: Teaching Models Plausible and Actionable Explanations". Our paper proposed a novel training regime for neural networks that induces greater explanatory capacity and improved adversarial robustness. The code is structured as a Julia package and was used to compute the experimental results presented in the paper in an aggregated form. The granular and aggregated results are also made available below. More detailed descriptions of the code can be found in the corresponding GH repository linked below. The corresponding paper will be published in late March, 2026: https://satml.org/