AI Paper Atlas

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

Rethinking Importance Weighting for Deep Learning under Distribution Shift

Tongtong Fang, Nan Lu, Gang Niu, Masashi Sugiyama

NeurIPS 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 12 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 a feature extractor to IW (SIW-L vs IW) (ablation)importance weighting · CIFAR-10 0.4 symmetric noiseaccuracy (%): 44.31 -> 76.39Improved
make weights dynamic (DIW1-L vs SIW-L) (ablation)static IW with FE · CIFAR-10 0.4 symmetric noiseaccuracy (%): 76.39 -> 75.27Mixed
make weights dynamic (DIW1-F vs SIW-F) (ablation)static IW with FE · CIFAR-10 0.4 symmetric noiseaccuracy (%): 65.58 -> 78.23Improved
update the feature extractor during training (DIW2-L vs DIW1-L) (ablation)DIW · CIFAR-10 0.4 symmetric noiseaccuracy (%): 75.27 -> 80.40Improved
pretrain FE for -F variant (DIW3-F vs DIW2-F) (ablation)DIW-F · CIFAR-10 0.4 symmetric noiseaccuracy (%): 77.48 -> 78.75Improved
pretrain FE for -L variant (DIW3-L vs DIW2-L) (ablation)DIW-L · CIFAR-10 0.4 symmetric noiseaccuracy (%): 80.40 -> 80.10No clear effect
DIW vs Reweight under label noiseDIW2-L · CIFAR-100 0.4 symmetric noiseaccuracy (%): 42.15 -> 53.66Improved
DIW vs best baseline under class-prior shift rho=100DIW · Fashion-MNISTaccuracy (%): 83.48 -> 84.02Improved
DIW vs IW weight estimation under class-prior shift rho=100DIW · Fashion-MNISTMAE to true weights: 1.10 -> 0.45Improved
DIW vs Reweight under pair-flip noise 0.3DIW · Fashion-MNISTaccuracy (%): 88.74 -> 88.19Hurt

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