AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Generalized Zero-Shot Learning via Over-Complete Distribution

Rohit Keshari, Richa Singh, Mayank Vatsa

CVPR 2020 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Generative feature synthesis for zero-shot · 144 citations (OpenAlex)

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 OCD to OBTL training (ablation)OBTL classifier · AWA2, SUN, and CUB PSclassification accuracy (%): 65.8 -> 70.9; 56.4 -> 62; 54.5 -> 60.5Improved
Add OCD to CL training (ablation)CL classifier · AWA2, SUN, and CUB PSclassification accuracy (%): 65.3 -> 66.5; 56.2 -> 57.6; 53.7 -> 56.8Improved
Add CL to OCD+OBTL (ablation)OCD-CVAE · AWA2, SUN, and CUB PSclassification accuracy (%): 70.9 -> 71.3; 62 -> 62.1; 60.5 -> 60.9Improved
Add OBTL to OCD+CL (ablation)OCD-CVAE · AWA2, SUN, and CUB PSclassification accuracy (%): 66.5 -> 71.3; 57.6 -> 62.1; 56.8 -> 60.9Improved
Use only OBTL or only CL, without OCD (ablation)OCD-CVAE · AWA2, SUN, and CUB PSclassification accuracy (%): 65.8, 56.4, 54.5; 65.3, 56.2, 53.7Hurt
Generate OCD and train with the proposed frameworkZero-shot classifier · AWA2 PSclassification accuracy (%): 69.5 -> 71.3Improved
Generate OCD and train with the proposed frameworkZero-shot classifier · CUB PSclassification accuracy (%): 59.6 -> 60.3Improved
Generate OCD and train with the proposed frameworkZero-shot classifier · SUN PSclassification accuracy (%): 63.4 -> 63.5No clear effect
Use the proposed OCD-CVAE frameworkGZSL classifier · AWA2 GZSLharmonic mean H (%): 63.9 -> 65.7Improved
Use the proposed OCD-CVAE frameworkGZSL classifier · CUB GZSLharmonic mean H (%): 52.4 -> 51.3Hurt
Use the proposed OCD-CVAE frameworkGZSL classifier · SUN GZSLharmonic mean H (%): 53.4 -> 43.8Hurt
Generate 400 to 600 OCD samples for trainingAWA2 regressor · AWA2 PSclassification accuracyImproved

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