Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks
Uddeshya Upadhyay, Shyamgopal Karthik, Yanbei Chen, Massimiliano Mancini, Zeynep Akata
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 identity mapping loss (ablation) | BayesCap · BSD100 super-resolution | UCE: 0.34 -> 0.028 | Improved |
| add identity mapping loss (ablation) | BayesCap · BSD100 super-resolution | C.Coeff: 0.23 -> 0.45 | Improved |
| use BayesCap instead of Generalized Gaussian scratch model | SRGAN · BSD100 super-resolution | UCE: 0.033 -> 0.028 | Improved |
| use BayesCap instead of Generalized Gaussian scratch model | SRGAN · BSD100 super-resolution | PSNR: 24.78 -> 25.16 | Improved |
| use BayesCap instead of Scratch trained from scratch | SRGAN · BSD100 super-resolution | PSNR: 24.39 -> 25.16 | Improved |
| use BayesCap instead of post-hoc variance-scaled Scratch | SRGAN · BSD100 super-resolution | UCE: 0.036 -> 0.028 | Improved |
| use BayesCap instead of Scratch | SRGAN · BSD100 super-resolution | ECE: 1.26 -> 0.83 | Improved |
| use BayesCap instead of dropout (DO) | SRGAN · BSD100 super-resolution | NLL: 0.47 -> 0.21 | Improved |
Later papers that compared with this method
1 later accepted paper reports BayesCap as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- CUPID: A Plug-in Framework for Joint Aleatoric and Epistemic Uncertainty Estimation with a Single Model (ICLR 2026 · better on 2 of 2 shared results)
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
- Bayesian Layers: A Module for Neural Network Uncertainty (NeurIPS 2019 · similarity 0.84)
- Bayesian Uncertainty Estimation for Batch Normalized Deep Networks (ICML 2018 · similarity 0.83)
- Make Sure You're Unsure: A Framework for Verifying Probabilistic Specifications (NeurIPS 2021 · similarity 0.83)
- Sketched Lanczos uncertainty score: a low-memory summary of the Fisher information (NeurIPS 2024 · similarity 0.82)
- Variational Bayesian Last Layers (ICLR 2024 · similarity 0.82)
- Make Me a BNN: A Simple Strategy for Estimating Bayesian Uncertainty from Pre-trained Models (CVPR 2024 · similarity 0.82)
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.
