AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Adaptive wavelet distillation from neural networks through interpretations

Wooseok Ha, Chandan Singh, Francois Lanusse, Srigokul Upadhyayula, Bin Yu

NeurIPS 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 4 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
distill LSTM into adaptive wavelet instead of fixed DB5 waveletwavelet model on LSTM · CME molecular partner predictionR^2: 0.197 -> 0.262Improved
AWD wavelet model versus the original LSTMLSTM · CME molecular partner predictionR^2: 0.237 -> 0.262Improved
use adaptive wavelet instead of DB5 for compressionwavelet model · CME molecular partner predictioncompression factor: 0.704 -> 0.574Improved
replace LSTM with AWD wavelet model at test timeLSTM · CME molecular partner predictioncomputation time: 0.0449s -> 0.0002sImproved
AWD wavelet filters instead of DB5 wavelet for peak findingpeak-counting predictor · Cosmological parameter predictionRMSE x 10^-2: 1.569 -> 1.029Improved
AWD filters instead of Laplace filterpeak-counting predictor · Cosmological parameter predictionRMSE x 10^-2: 1.369 -> 1.029Improved
AWD filters instead of Roberts-Cross filterpeak-counting predictor · Cosmological parameter predictionRMSE x 10^-2: 1.259 -> 1.029Improved
AWD interpretable model instead of the full ResNetResNet18 · Cosmological parameter predictionRMSE x 10^-2: 1.156 -> 1.029Improved
AWD instead of DB5 wavelet for compressionwavelet model · Cosmological parameter predictioncompression rate: 0.620 -> 0.610Improved
vary interpretation penalty gamma (too small or too large) (ablation)AWD · Simulated DB5 recoverydistance to groundtruth waveletMixed
initialize lowpass filter at Sym 5 versus DB5+noise, varying sparsity penalty lambda (ablation)AWD · Simulated DB5 recoverydistance to groundtruth waveletMixed
initialize lowpass filter at Coif 2, varying sparsity penalty lambda (ablation)AWD · Simulated DB5 recoverydistance to groundtruth waveletMixed

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