Research map / Architectures and efficiency / RNNs and normalization
Shapley Explanation Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use Shallow ShapNet instead of post-hoc approximations | Shapley explanation computation · Synthetic regression | normalized L1 Shapley error: 4.19e-08 | Helped |
| Use Deep ShapNet for intrinsic explanations | Shapley explanation computation · Yeast | normalized L1 Shapley error: 0.0504 | Mixed |
| Use Deep ShapNet instead of the comparable CNN | image classification · Cifar-10 | accuracy: 0.7996 -> 0.8206 | Helped |
| Use Deep ShapNet instead of the state-of-the-art model | image classification · FashionMNIST | accuracy: 0.9195 -> 0.9691 | Hurt |
| Apply ℓ1 explanation regularization (ablation) | Deep ShapNet · Yeast | sparsity of SHAP values: 0.00275 -> 0.00425 | Helped |
| Apply ℓ1 explanation regularization (ablation) | Deep ShapNet · Breast Cancer Wisconsin | sparsity of SHAP values: 0.209 -> 0.841 | Helped |
| Apply ℓ1 explanation regularization (ablation) | Deep ShapNet · Yeast | accuracy: 0.587 -> 0.592 | Helped |
| Apply ℓ1 explanation regularization (ablation) | Deep ShapNet · Breast Cancer Wisconsin | accuracy: 0.960 -> 0.960 | No clear effect |
| Apply ℓ∞ explanation regularization (ablation) | Deep ShapNet · Yeast | coefficient of variation for abs. SHAP: 1.05 -> 0.768 | Helped |
| Apply ℓ∞ explanation regularization (ablation) | Deep ShapNet · Breast Cancer Wisconsin | coefficient of variation for abs. SHAP: 2.04 -> 1.28 | Helped |
| Apply ℓ1 regularization on all inter-layer representations rather than only the output (ablation) | Deep ShapNet · MNIST test set | accuracy retained after pruning | Helped |
Closest papers by meaning
- SHAP Meets Tensor Networks: Provably Tractable Explanations with Parallelism (NeurIPS 2025 · similarity 0.88)
- shapiq: Shapley Interactions for Machine Learning (NeurIPS 2024 · similarity 0.87)
- SHAP values via sparse Fourier representation (NeurIPS 2025 · similarity 0.87)
- Prediction via Shapley Value Regression (ICML 2025 · similarity 0.87)
- InstaSHAP: Interpretable Additive Models Explain Shapley Values Instantly (ICLR 2025 · similarity 0.86)
- Data Shapley in One Training Run (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.