Research map / Probabilistic and scientific modeling / Variational and generative sampling
Variational Inference for Infinitely Deep Neural Networks
Achille Nazaret, David Blei
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 a posterior over truncation depths and combine predictions across those depths | UDN with fully connected networks · Spiral classification D(ω), test accuracy averaged over ω | accuracy (%): 90.5 -> 91.7 | Improved |
| Use a posterior over truncation depths and combine predictions across those depths | UDN with ResNet · CIFAR-10 test | accuracy: 94.0 -> 94.4 | Improved |
| Increase spiral rotation complexity | UDN posterior · Spiral classification D(ω) | posterior truncation depth | Mixed |
| Use a harder class pair (cat vs dog) instead of an easier pair (deer vs car) | UDN posterior · CIFAR-10 two-class subsamples | posterior truncation depth | Improved |
| Use the full CIFAR-10 task instead of the two-class subsamples | UDN posterior · CIFAR-10 | posterior truncation depth | Improved |
| Combine predictions from multiple depths rather than use an individual finite-depth network | DUN-10 · Spiral classification D(ω), test accuracy averaged over ω | accuracy (%): 89.3 -> 90.5 | Improved |
Closest papers by meaning
- Liberty or Depth: Deep Bayesian Neural Nets Do Not Need Complex Weight Posterior Approximations (NeurIPS 2020 · similarity 0.89)
- On the Expressiveness of Approximate Inference in Bayesian Neural Networks (NeurIPS 2020 · similarity 0.89)
- Radial and Directional Posteriors for Bayesian Deep Learning (AAAI 2020 · similarity 0.89)
- VIKING: Deep variational inference with stochastic projections (NeurIPS 2025 · similarity 0.88)
- Neural Variational Inference and Learning in Undirected Graphical Models (NeurIPS 2017 · similarity 0.88)
- Globally Convergent Variational Inference (NeurIPS 2024 · similarity 0.87)
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.
