Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning Discrete Structured Representations by Adversarially Maximizing Mutual Information
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 AMMI instead of BMSH or DVQ at 128 bits | Document hashing · TMC, NG20, Reuters | top-100 precision: TMC: 76.27 vs 76.24; NG20: 66.18 vs 62.10; Reuters: 86.02 vs 86.27 | Mixed |
| Use AMMI instead of the strongest prior baseline | Document hashing · NG20, 64 bits | top-100 precision: 58.77 -> 63.98 | Helped |
| Use AMMI instead of the strongest prior baseline | Document hashing · Reuters, 16 bits | top-100 precision: 79.54 -> 81.73 | Helped |
| Use AMMI instead of the strongest prior baseline | Document hashing · Reuters, 32 bits | top-100 precision: 83.43 -> 84.46 | Helped |
| Use AMMI instead of the strongest prior baseline | Document hashing · Reuters, 64 bits | top-100 precision: 83.73 -> 85.06 | Helped |
| Use AMMI instead of the strongest prior baseline | Document hashing · TMC, 16 bits | top-100 precision: 71.47 -> 70.96 | Hurt |
| Use AMMI instead of the strongest prior baseline | Document hashing · TMC, 32 bits | top-100 precision: 74.81 -> 74.16 | Hurt |
| Use AMMI instead of the strongest prior baseline | Document hashing · NG20, 16 bits | top-100 precision: 58.12 -> 55.18 | Hurt |
| Use AMMI instead of the strongest prior baseline | Document hashing · NG20, 32 bits | top-100 precision: 61.00 -> 59.56 | Hurt |
| Use AMMI instead of the strongest prior baseline | Document hashing · Predictive document hashing | top-100 precision: 76.80 -> 79.14 | Helped |
| Set the prior Markov order strictly higher than the encoder order (ablation) | AMMI · Reuters validation, 16 bits | top-100 precision | Helped |
| Use a higher-order variational prior to estimate encoder entropy (ablation) | AMMI · Reuters, 16-bit prior-order experiment | cross entropy | Helped |
Closest papers by meaning
- Learning deep representations by mutual information estimation and maximization (ICLR 2019 · similarity 0.86)
- On Mutual Information Maximization for Representation Learning (ICLR 2020 · similarity 0.86)
- The Information Sieve (ICML 2016 · similarity 0.85)
- Mutual Information Neural Estimation (ICML 2018 · similarity 0.85)
- Representation Unlearning: Forgetting through Information Compression (ICML 2026 · similarity 0.85)
- Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning (NeurIPS 2025 · similarity 0.83)
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.