AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Attribution Preservation in Network Compression for Reliable Network Interpretation

Geondo Park, June Yong Yang, Sung Ju Hwang, Eunho Yang

NeurIPS 2020 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 6 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
Add stochastic sensitivity-weighted attribution matching during VGG16/8 distillationVGG16/8 distilled from VGG16 · PASCAL VOC 2012 validationAUC: 78.74 -> 88.06Improved
Add stochastic sensitivity-weighted attribution matching during VGG16/8 distillationVGG16/8 distilled from VGG16 · PASCAL VOC 2012 validationmAP: 76.91 -> 80.86Improved
Replace naive fine-tuning with stochastic sensitivity-weighted matching after unstructured pruningVGG16 · PASCAL VOC 2012 validationAUC: 84.13 -> 88.45Improved
Replace naive fine-tuning with stochastic sensitivity-weighted matching after structured pruningVGG16 · PASCAL VOC 2012 validationAUC: 79.98 -> 88.34Improved
Add stochastic sensitivity-weighted matching after 60% unstructured pruningVGG16 · ImageNet segmentation subsetAUC: 76.01 -> 80.88Improved
Add stochastic sensitivity-weighted matching after structured pruningResNet34 · ImageNet segmentation subsetAUC: 81.20 -> 84.65Improved
Use sensitivity-weighted matching rather than equally weighted activation matching (ablation)VGG16/8 distillation · PASCAL VOC 2012 validationAUC: 81.99 -> 87.88Improved
Add stochastic channel masking to sensitivity-weighted matching (ablation)VGG16/8 distillation · PASCAL VOC 2012 validationAUC: 87.88 -> 88.06Improved
Use sensitivity-weighted matching rather than equally weighted matching (ablation)VGG16/8 distillation · PASCAL VOC 2012 validationCosine similarity: 0.652 -> 0.813Improved
Use stochastic sensitivity-weighted Grad-CAM matchingVGG16/8 distillation · PASCAL VOC 2012 validationAUC for Excitation Backprop: 76.31 -> 82.31Improved
Use stochastic sensitivity-weighted Grad-CAM matchingVGG16/8 distillation · PASCAL VOC 2012 validationAUC for LRP: 79.60 -> 82.46Improved
Use stochastic sensitivity-weighted Grad-CAM matchingVGG16/8 distillation · PASCAL VOC 2012 validationPoint accuracy for RAP: 56.87 -> 65.66Improved

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