Research map / Probabilistic and scientific modeling / Variational and generative sampling
Faster Attend-Infer-Repeat with Tractable Probabilistic Models
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 SuPAIR instead of AIR | SuPAIR · Multi-MNIST | count accuracy: 94.7% -> 98.3% | Helped |
| use SuPAIR instead of AIR | SuPAIR · Sprites | count accuracy: 95.9% -> 99.2% | Helped |
| use SuPAIR instead of AIR on Noisy-MNIST | SuPAIR · Noisy-MNIST | count accuracy: 0.0% -> 93.8% | Helped |
| use SuPAIR instead of AIR on Grid-MNIST | SuPAIR · Grid-MNIST | count accuracy: 0.0% -> 97.5% | Helped |
| remove the learned background model (SuPAIR w/o bg) (ablation) | SuPAIR · Noisy and Grid MNIST | count accuracy | Hurt |
| replace baseline networks with enumeration of object count N (ablation) | SuPAIR · MNIST | training epoch time | Helped |
| use SuPAIR tractable appearance models instead of AIR VAE | SuPAIR · Multi-MNIST | training time to convergence | Helped |
| use Poon & Domingos SPN structure instead of random RAT-SPN (ablation) | SuPAIR · image datasets | speed and accuracy | No clear effect |
| use SuPAIR instead of AIR | SuPAIR · Parameter count | total parameters: 6,104,117 -> 3,213,145 | Helped |
| add Gaussian noise or grid background to Multi-MNIST | SuPAIR · Multi-MNIST | count accuracy: 98.3% -> 93.8% | Hurt |
Closest papers by meaning
- Data-efficient Large Vision Models through Sequential Autoregression (ICML 2024 · similarity 0.86)
- CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026 · similarity 0.85)
- Zero-Shot Scene Reconstruction from Single Images with Deep Prior Assembly (NeurIPS 2024 · similarity 0.85)
- BVS: Bayesian Visual Search with Multimodal Large Language Model for Fine-grained Perception (ICML 2026 · similarity 0.85)
- Scalable Pre-training of Large Autoregressive Image Models (ICML 2024 · similarity 0.84)
- Adaptive Visual Scene Understanding: Incremental Scene Graph Generation (NeurIPS 2024 · 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.