AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Less Is Better: Unweighted Data Subsampling via Influence Function

Zifeng Wang, Hong Zhu, Zhenhua Dong, Xiuqiang He, Shao-Lun Huang

AAAI 2020 · Learning theory and algorithms › Prediction and decision losses · approach: Active learning and subset selection · 40 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
Use Sigmoid influence sampling instead of full-set trainingLogistic regression · UCI Adultaverage logloss: 0.3383 -> 0.3382Improved
Use Sigmoid influence sampling instead of full-set trainingLogistic regression · cifar10average logloss: 0.6847 -> 0.6819Improved
Use Sigmoid influence sampling instead of full-set trainingLogistic regression · Avazu-siteaverage logloss: 0.4499 -> 0.4486Improved
Use Linear influence sampling instead of full-set trainingLogistic regression · Criteo1%average logloss: 0.4763 -> 0.4755Improved
Use Linear influence sampling instead of full-set trainingLogistic regression · Companyaverage logloss: 0.1955 -> 0.1952Improved
Use Sigmoid influence sampling instead of DropoutLogistic regression · UCI breast-canceraverage logloss: 0.0785 -> 0.0803Hurt
Use Sigmoid influence sampling instead of DropoutLogistic regression · diabetesaverage logloss: 0.5083 -> 0.5068Improved
Use Sigmoid influence sampling instead of DropoutLogistic regression · News20average logloss: 0.5072 -> 0.5075Hurt
Use Linear influence sampling instead of DropoutLogistic regression · MNISTaverage logloss: 0.0223 -> 0.0245Hurt
Use Linear influence sampling instead of DropoutLogistic regression · Covertypeaverage logloss: 0.7745 -> 0.6872Improved
Use probabilistic influence sampling rather than deterministic Dropout (ablation)UIDS · out-of-sample test across datasetsloglossMixed
Use Linear sampling rather than Sigmoid sampling (ablation)UIDS · Criteo1%average logloss: 0.4756 -> 0.4755Improved

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