Research map / Architectures and efficiency / Attention and transformers
Learning to Generate with Memory
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 |
|---|---|---|---|
| add external memory to VAE (ablation) | VAE · MNIST | log-likelihood: -85.67 -> -84.41 | Helped |
| add external memory to IWAE-50 (ablation) | IWAE-50 · MNIST | log-likelihood: -83.67 -> -82.84 | Helped |
| add external memory to IWAE-5 (ablation) | IWAE-5 · OCR-letters | log-likelihood: -28.69 -> -27.65 | Helped |
| add external memory to VAE (ablation) | VAE · OCR-letters | log-likelihood: -30.09 -> -29.09 | Helped |
| enlarge VAE to 530-530-100 to match parameters instead of adding memory (ablation) | VAE · MNIST | log-likelihood (1 sample): -85.67 -> -85.69 | No clear effect |
| remove local reconstruction error regularizer (ablation) | MEM-VAE · MNIST | log-likelihood: -84.41 -> -84.44 | No clear effect |
| add local reconstruction regularizer to VAE without memory (ablation) | VAE · MNIST | log-likelihood: -85.67 -> -85.68 | No clear effect |
| add memory to VAE (ablation) | VAE · Frey faces batch 100 | log density: 1308 -> 1330 | Helped |
| add memory to VAE (ablation) | VAE · Frey faces batch 10 | log density: 1055 -> 1240 | Helped |
| add memory to VAE (ablation) | VAE · MNIST RECT-12 imputation | MSE: 0.1403 -> 0.1362 | Helped |
| add memory to VAE (ablation) | VAE · MNIST RAND-0.6 imputation | MSE: 0.0194 -> 0.0187 | Helped |
| add memory to VAE (ablation) | VAE · MNIST HALF imputation | MSE: 0.0550 -> 0.0539 | Helped |
Closest papers by meaning
- Variational Memory Addressing in Generative Models (NeurIPS 2017 · similarity 0.86)
- Generative Models as a Data Source for Multiview Representation Learning (ICLR 2022 · similarity 0.86)
- An Architecture for Deep, Hierarchical Generative Models (NeurIPS 2016 · similarity 0.85)
- Task-Free Continual Generation and Representation Learning via Dynamic Expansionable Memory Cluster (AAAI 2024 · similarity 0.85)
- Can Generative Models Improve Self-Supervised Representation Learning? (AAAI 2025 · similarity 0.85)
- Towards flexible perception with visual memory (ICML 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.