Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Partially Exchangeable Networks and Architectures for Learning Summary Statistics in Approximate Bayesian Computation
Samuel Wiqvist, Pierre-Alexandre Mattei, Umberto Picchini, Jes Frellsen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 PEN-0 instead of MLP pre | ABC summary learning for α-stable · α-stable, 5·10^5 training observations | RMSE: 0.07 vs 0.05 | Improved |
| Use PEN-0 instead of MLP pre | ABC summary learning for α-stable · α-stable, 10^3 training observations | RMSE: 0.40 vs 0.07 | Improved |
| Use empirical distribution function as MLP input rather than raw observations | ABC summary learning for α-stable · α-stable | RMSE: 0.21 vs 0.07 | Improved |
| Use PEN-2 instead of MLP | ABC summary learning for AR(2) · AR(2) | Wasserstein distance | Improved |
| Use fewer training observations with PEN-2 rather than MLP | ABC summary learning for AR(2) · AR(2) | Wasserstein distance: 10^3; 10^5 | Improved |
| Use PEN-10 instead of MLP | ABC summary learning for MA(2) with observation noise · MA(2) | Wasserstein distance | Improved |
| Use PEN-0 instead of MLP for static-model summary learning | ABC summary learning for g-and-k · g-and-k | Wasserstein distance | Improved |
| Use MLP pre rather than MLP on raw observations | ABC summary learning for g-and-k · g-and-k | Wasserstein distance | Improved |
| Use PEN-0 instead of MLP pre with less training data | ABC summary learning for g-and-k · g-and-k | Wasserstein distance | Improved |
| Reduce training observations for MLP | ABC summary learning for AR(2) · AR(2) | Wasserstein distance: 10^3; 10^4 | Hurt |
| Reduce training observations for PEN-2 | ABC summary learning for AR(2) · AR(2) | Wasserstein distance | No clear effect |
Closest papers by meaning
- BRUNO: A Deep Recurrent Model for Exchangeable Data (NeurIPS 2018 · similarity 0.87)
- Partially Stochastic Infinitely Deep Bayesian Neural Networks (ICML 2024 · similarity 0.86)
- Replica-Exchange Nos\'e-Hoover Dynamics for Bayesian Learning on Large Datasets (NeurIPS 2020 · similarity 0.84)
- Architectural and Inferential Inductive Biases for Exchangeable Sequence Modeling (NeurIPS 2025 · similarity 0.84)
- A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors (ICLR 2024 · similarity 0.84)
- Scalable Bayesian Learning with posteriors (ICLR 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.
