AI Paper Atlas

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

Identifying Useful Learnwares for Heterogeneous Label Spaces

Lan-Zhe Guo, Zhi Zhou, Yu-Feng Li, Zhi-Hua Zhou

ICML 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias

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
class-wise specification and class-wise identification (vs. RKME-basic)learnware identification · Homogeneous and heterogeneous label spaces, Pre@kPre@1: e.g. 95.15 vs 54.54 (direct use); 59.09 vs 36.36 (heterogeneous)Improved
class-wise specification (vs. RKME-basic) for trained-model correlationspecification similarity ranking · 22 user tasks, Kendall's tauKendall tau: outperforms on 20 of 22 datasetsImproved
reuse small model selected via specification instead of fine-tuning generic big model DenseNet201model reuse on user tasks · 22 user tasks, average accuracyaccuracy: 43.54 vs 32.99Improved
class-wise specification selection vs. brute-force fine-tuning of all modelslearnware identification cost · Timingspeedup: at least thousand-fold (fine-tuning ~48s vs similarity ~0.003s)Improved

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