Reproduce “Visualizing Deep Neural Network Decisions: Prediction Difference Analysis” 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
Luisa M Zintgraf, Taco S Cohen, Tameem Adel, Max Welling
The authors present prediction-difference visualizations that explain classifier decisions by replacing image patches with plausible alternatives and measuring how class predictions change. Conditioning replacements on neighboring pixels yields more object-focused maps than marginal sampling in two ImageNet examples. The maps show evidence for and against a class, but do not establish ground-truth explanations or recover the model’s internal reasoning.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.