Research map / Architectures and efficiency / RNNs and normalization
Scalable Model Compression by Entropy Penalized Reparameterization
Deniz Oktay, Johannes Ballé, Saurabh Singh, Abhinav Shrivastava
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 entropy-penalized reparameterization instead of Sparse Variational Dropout | LeNet300-100 · MNIST | size: 9.38 KB (113x) -> 8.56 KB (124x) | Improved |
| Use entropy-penalized reparameterization instead of Minimal Random Code Learning | LeNet5-Caffe · MNIST | Error (Top-1): 1.0% -> 0.9% | Improved |
| Replace scalar quantization with a DFT reparameterization | VGG-16 · CIFAR-10 | size: 168 KB (452x) -> 101 KB (590x) | Improved |
| Replace scalar quantization with a DFT reparameterization (ablation) | ResNet-20-4 · CIFAR-10 | Error (Top-1): 10.3% -> 8.8% | Improved |
| Replace scalar quantization with a DFT reparameterization (ablation) | ResNet-18 · ImageNet | size: 2.78 MB (17x) -> 1.97 MB (24x) | Improved |
| Replace scalar quantization with a DFT reparameterization (ablation) | ResNet-50 · ImageNet | Error (Top-1): 26.5% -> 26.0% | Improved |
| Use a DFT transform rather than a random fixed orthogonal matrix (ablation) | Convolution kernels in ResNet-18 and ResNet-20-4 · ImageNet and CIFAR-10 | compression performance | Improved |
| Use a DFT transform rather than scalar quantization (ablation) | CIFAR-10 convolution layers · CIFAR-10 | compression performance | Improved |
Closest papers by meaning
- On Compressing Deep Models by Low Rank and Sparse Decomposition (CVPR 2017 · similarity 0.89)
- Rate Distortion For Model Compression:From Theory To Practice (ICML 2019 · similarity 0.89)
- Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters (ICLR 2019 · similarity 0.88)
- Compressing Neural Networks: Towards Determining the Optimal Layer-wise Decomposition (NeurIPS 2021 · similarity 0.88)
- MCNC: Manifold-Constrained Reparameterization for Neural Compression (ICLR 2025 · similarity 0.88)
- Model Preserving Compression for Neural Networks (NeurIPS 2022 · 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.
