Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Conditional Neural Processes
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 CNP instead of GP or kNN | CelebA image completion · CelebA with 10 or 100 context points | mean squared error: 0.215, 0.247 -> 0.039; 0.052, 0.137 -> 0.016 (random context) | Helped |
| Use CNP instead of GP or kNN | CelebA image completion · CelebA with 1000 random context points | mean squared error: 0.007, 0.001 -> 0.009 | Mixed |
| Provide context points in top-to-bottom order rather than randomly | CNP · CelebA image completion | mean squared error: 0.039 -> 0.057; 0.016 -> 0.047; 0.009 -> 0.021 | Hurt |
| Use CNP instead of MANN | Omniglot one-shot classification · Omniglot 5-way, 1-shot | accuracy: 82.8% -> 95.3% | Helped |
| Use CNP instead of Matching Networks | Omniglot one-shot classification · Omniglot | accuracy: 98.1% -> 95.3%; 98.9% -> 98.5%; 93.8% -> 89.9%; 98.5% -> 96.8% | Mixed |
| Add a latent variable to the decoder to model global uncertainty (ablation) | CNP · MNIST and CelebA image completion | coherent sample generation | Helped |
| Increase the number of context points | CNP · 1-D function regression | prediction accuracy | Helped |
| Choose observations by highest predictive variance rather than randomly | CNP · MNIST image completion | prediction accuracy | Helped |
| Use a factored output model without latent variables (ablation) | CNP · MNIST image completion with complete observations | pixel reconstruction | Hurt |
Closest papers by meaning
- Flow Matching Neural Processes (NeurIPS 2025 · similarity 0.86)
- Autoencoding Conditional Neural Processes for Representation Learning (ICML 2024 · similarity 0.86)
- Rényi Neural Processes (ICML 2025 · similarity 0.86)
- Neural Conditional Probability for Uncertainty Quantification (NeurIPS 2024 · similarity 0.85)
- Spectral Convolutional Conditional Neural Process (NeurIPS 2025 · similarity 0.85)
- Bayesian Optimization via Continual Variational Last Layer Training (ICLR 2025 · similarity 0.85)
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.