Reproduce “VCNet: A self-explaining model for realistic counterfactual generation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Victor Guyomard, Françoise Fessant, Thomas Guyet, Tassadit Bouadi, Alexandre Termier
VCNet addresses counterfactual recourse by jointly training a predictor and conditional variational autoencoder (cVAE) that reuses an input’s latent representation while changing its class condition. This avoids per-example optimization. Across six binary-classification tabular datasets, it achieves better prediction gain and class-proximity scores than CounterNet. On Adult, its class-proximity score is 0.04 ± 0.11, compared with 0.31 ± 0.28; validity is not perfect on two datasets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.