Research map / Probabilistic and scientific modeling / Molecules and proteins
Explaining A Black-box By Using A Deep Variational Information Bottleneck Approach
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 VIBI explanations instead of L2X explanations | LSTM sentiment predictions · IMDB | human sentiment inference accuracy: 35.6% -> 44.7% | Helped |
| Use VIBI explanations instead of Saliency explanations | CNN digit predictions · MNIST | human explanation score: 3.448 -> 3.526 | Helped |
| Add the information-bottleneck compressiveness term (β>0) to the explanation objective (ablation) | VIBI · IMDB and MNIST | rationale fidelity: IMDB word chunks, k=5: 0.638 -> 0.666; MNIST 4×4 chunks, k=4: 0.650 -> 0.775 | Helped |
| Increase β from 0 to 0.001 (ablation) | VIBI · MNIST (4×4 chunks, k=4) | approximator fidelity accuracy: 0.953 -> 0.922 | Hurt |
| Increase β from 0 to 0.1 (ablation) | VIBI · MNIST (4×4 chunks, k=4) | rationale fidelity accuracy: 0.650 -> 0.775 | Helped |
| Increase β from 0 to 0.01 (ablation) | VIBI · IMDB (word chunks, k=5) | rationale fidelity accuracy: 0.638 -> 0.666 | Helped |
Closest papers by meaning
- Beyond Concept Bottleneck Models: How to Make Black Boxes Intervenable? (NeurIPS 2024 · similarity 0.85)
- Gnothi Seauton: Empowering Faithful Self-Interpretability in Black-Box Transformers (ICLR 2025 · similarity 0.84)
- From Black-box to Causal-box: Towards Building More Interpretable Models (NeurIPS 2025 · similarity 0.84)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.84)
- Local vs. Global Interpretability: A Computational Complexity Perspective (ICML 2024 · similarity 0.84)
- Sample-efficient Learning of Concepts with Theoretical Guarantees: from Data to Concepts without Interventions (NeurIPS 2025 · similarity 0.84)
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.