AI Paper Atlas

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

Learnability with Indirect Supervision Signals

Kaifu Wang, Qiang Ning, Dan Roth

NeurIPS 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 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
Use indirect supervision with a transition class whose induced label distributions are separatedMulticlass classification from indirect supervision · Realizable multiclass classificationlearnabilityImproved
Restrict the transition class using prior information that separates label-induced annotation distributionsMulticlass classification from indirect supervision · Realizable multiclass classificationidentifiability: η ≥ γ > 0Improved
Use a transition class with no positive separationMulticlass classification from indirect supervision · Realizable multiclass classificationidentifiability: lim_k η^(k)=0Hurt
Use concentration sets that distinguish the annotation distributions for each labelMulticlass classification from indirect supervision · Concentration conditionseparation degree: γ ≥ 2γ_C^2Improved
Use the transition-independent loss 1{o not in S_h(x)} under the concentration conditionERM for indirect supervision · Concentration conditionidentifiability: η ≥ γ_CImproved
Combine two annotation sources without constraints on their joint transitionJoint supervision · Pairwise label separationseparation degree: γ_i→j ≤ λγ_i→j1 + (1−λ)γ_i→j2Mixed
Distinguish the identities of annotation sources in joint supervisionJoint supervision · Pairwise label separationseparation degree: γ_i→j = λγ_i→j1 + (1−λ)γ_i→j2Improved
Mix annotation sources without retaining source identityJoint supervision · Binary classification with two annotatorsseparation: λ=1/2Hurt
Impose prior knowledge that one annotator has better quality than the otherJoint supervision with two noisy annotators · Binary classificationlearnability: η_1−η_2≤γ<0Improved

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