AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Pre-Train Your Loss: Easy Bayesian Transfer Learning with Informative Priors

Ravid Shwartz-Ziv, Micah Goldblum, Hossein Souri, Sanyam Kapoor, Chen Zhu, Yann LeCun, Andrew Gordon Wilson

NeurIPS 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervision in supervised pipelines · 8 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use an informative learned prior instead of a non-learned priorBayesian transfer with SimCLR ResNet-50 · CIFAR-10, 10 samplestest error: 86.1± 3.9 -> 73.8± 3.2Improved
Use Bayesian inference with the learned prior rather than SGDSimCLR ResNet-50 transfer · CIFAR-100, 10 samplestest error: 90.9± 2.8 -> 81.9± 3.7Improved
Use a learned SSL prior rather than a learned supervised priorBayesian segmentation · PASCAL VOC 2012 val, ResNet-50Mean-IoU: 73.72 -> 74.15Improved
Use a learned SSL prior rather than a learned supervised priorBayesian segmentation · Cityscapes val, ResNet-50Mean-IoU: 76.09 -> 76.52Improved
Replace the learned covariance with an isotropic covariance while retaining the learned mean (ablation)Bayesian transfer with learned prior · CIFAR-100performanceHurt
Remove the low-rank covariance component (ablation)SWAG learned prior · CIFAR-10 and CIFAR-100performanceHurt
Use rank-5 rather than rank-0 low-rank covariance (ablation)SWAG learned prior · CIFAR-10performanceImproved
Increase the low-rank covariance rank through 7 (ablation)SWAG learned prior · CIFAR-100performanceImproved
Rescale the learned prior covariance to be more diffuse for the downstream task (ablation)Bayesian transfer with learned prior · CIFAR-10accuracyImproved
Use Bayesian model averaging rather than MAP estimation (ablation)Transfer with learned prior · Downstream classification tasksperformanceImproved
Use a learned prior rather than standard pre-trained initializationSGD transfer learning · Semantic segmentationMean-IoU: 73.17 -> 73.48Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy