Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Compression for Deep Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 group-level sparsity with a normal-Jeffreys prior and posterior-based weight precision | Bayesian Compression · LeNet-5-Caffe | maximum compression rate: 365(1.0) -> 573(1.0) | Helped |
| Use group-level sparsity with a horseshoe prior and posterior-based weight precision | Bayesian Compression · LeNet-5-Caffe | maximum compression rate: 365(1.0) -> 771(1.0) | Helped |
| Use group-level sparsity with a normal-Jeffreys prior and posterior-based weight precision | Bayesian Compression · LeNet-300-100 | maximum compression rate: 113 (1.8) -> 58(1.8) | Hurt |
| Use group-level sparsity with a horseshoe prior and posterior-based weight precision | Bayesian Compression · VGG on CIFAR-10 | maximum compression rate: 95(8.6) vs 116(9.2) | Mixed |
| Use posterior uncertainties to set layer weight precision | Bayesian Compression · LeNet-5-Caffe | fast prediction compression rate: 228(1.0) -> 361(1.0) | Helped |
Closest papers by meaning
- Enhanced Bayesian Compression via Deep Reinforcement Learning (CVPR 2019 · similarity 0.89)
- Compression-aware Training of Deep Networks (NeurIPS 2017 · similarity 0.88)
- Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters (ICLR 2019 · similarity 0.88)
- Compression with Bayesian Implicit Neural Representations (NeurIPS 2023 · similarity 0.88)
- Effective Interplay between Sparsity and Quantization: From Theory to Practice (ICLR 2025 · similarity 0.88)
- Probabilistic Connection Importance Inference and Lossless Compression of Deep Neural Networks (ICLR 2020 · similarity 0.88)
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.