AI Paper Atlas

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

Recovery and Generalization in Over-Realized Dictionary Learning

Jeremias Sulam, Chong You, Zhihui Zhu

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory

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
increase dictionary size from p=70 to p'>70 (up to 500) during ODL+OMP learningOnline Dictionary Learning (ODL) with OMP · Synthetic 50x70 dictionary, k=3, 300 training samplestrain/test riskImproved
increase dictionary size from p=70 to p'>70 (up to 500) during ODL+OMP learningOnline Dictionary Learning (ODL) with OMP · Synthetic 50x70 dictionary, k=3, 300 training samplesrecovery distance to ground truthImproved
keep increasing dictionary size well beyond the ground-truth size with fixed 300 samplesODL over-realized dictionary · Synthetic 50x70 dictionary, k=3, 300 training samplesgeneralization errorHurt
increase number of training samples for a given over-realized dictionary sizeODL over-realized dictionary · Synthetic 50x70 dictionary, k=3risk and recovery errorImproved
distill over-realized dictionary to p atoms by keeping the p most frequently used atomsODL+OMP over-realized dictionary (100 atoms, 500 samples) · Synthetic 50x70 dictionary, k=3risk and dictionary errorImproved
over-realization plus distillation across sparsity, dimension and redundancy settingsODL+OMP · Synthetic dictionaries with varied parametersrisk and log10 dictionary errorImproved
use ODL with l1 regularizer (Lasso coding) instead of OMPover-realization plus distillation · Synthetic dictionariesdictionary recovery errorImproved
use batch K-SVD instead of ODLover-realization plus distillation · Synthetic dictionariesdictionary recovery errorImproved
add Gaussian noise (variance sigma^2) to training samplesover-realization and distillation · Synthetic dictionaries with noiserelative improvement in recovery over traditional learningMixed
retrain from initialization using only the atoms found by distillation (ablation)over-realized dictionary learning · Synthetic dictionariesrecovery performanceHurt

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