Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
BRUNO: A Deep Recurrent Model for Exchangeable Data
Iryna Korshunova, Jonas Degrave, Ferenc Huszar, Yarin Gal, Arthur Gretton, Joni Dambre
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 |
|---|---|---|---|
| add discriminative fine-tuning to BRUNO (ablation) | BRUNO · Omniglot few-shot classification | accuracy: 86.3% -> 97.1% | Improved |
| add discriminative fine-tuning to BRUNO (ablation) | BRUNO · Omniglot few-shot classification | accuracy: 95.6% -> 99.4% | Improved |
| add discriminative fine-tuning to BRUNO (ablation) | BRUNO · Omniglot few-shot classification | accuracy: 69.2% -> 91.3% | Improved |
| add discriminative fine-tuning to BRUNO (ablation) | BRUNO · Omniglot few-shot classification | accuracy: 87.7% -> 97.8% | Improved |
| use generative rather than discriminative training | BRUNO · Omniglot few-shot classification | accuracy: 86.3%, 95.6%, 69.2%, 87.7% vs 98.1%, 98.9%, 93.8%, 98.5% | Mixed |
| use weight normalization in Real NVP layers (ablation) | BRUNO with Gaussian processes · Fashion-MNIST training | training convergence | Improved |
| use Student-t processes instead of Gaussian processes (ablation) | BRUNO · training robustness | training convergence | Improved |
| add fully connected coupling layers on top of convolutional Real NVP (ablation) | BRUNO · Omniglot, Fashion-MNIST, and CIFAR-10 | model performance | Improved |
Later papers that compared with this method
2 later accepted papers report BRUNO as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- SurVAE Flows: Surjections to Bridge the Gap between VAEs and Flows (NeurIPS 2020 · better on 1 of 1 shared result)
- Exchangeable Generative Models with Flow Scans (AAAI 2020 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Partially Exchangeable Networks and Architectures for Learning Summary Statistics in Approximate Bayesian Computation (ICML 2019 · similarity 0.87)
- Architectural and Inferential Inductive Biases for Exchangeable Sequence Modeling (NeurIPS 2025 · similarity 0.85)
- Replica-Exchange Nos\'e-Hoover Dynamics for Bayesian Learning on Large Datasets (NeurIPS 2020 · similarity 0.85)
- Recurrent Kalman Networks: Factorized Inference in High-Dimensional Deep Feature Spaces (ICML 2019 · similarity 0.84)
- Scalable Bayesian Learning with posteriors (ICLR 2025 · similarity 0.83)
- VIKING: Deep variational inference with stochastic projections (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.
