AI Paper Atlas

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

Concept Distillation: Leveraging Human-Centered Explanations for Model Improvement

Avani Gupta, Saurabh Saini, P J Narayanan

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Post-hoc explanation analysis · 3 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 DINO teacher for CAV learning (concept distillation) (ablation)Concept loss with prototypes · ColorMNISTaccuracy (%): 26.97 -> 41.83Improved
add prototype-based intermediate-layer sensitivity (ablation)Concept distillation · ColorMNISTaccuracy (%): 30.94 -> 50.93Improved
add local explanation (RRR) loss to concept lossOurs · ColorMNISTaccuracy (%): 41.83 -> 50.93Improved
add local explanation (RRR) loss to concept lossOurs · DecoyMNISTaccuracy (%): 98.58 -> 98.98No clear effect
use concept loss vs CDEP baselineOurs+L · TextureMNISTaccuracy (%): 10.18 -> 56.57Improved
use concept sensitivity with intermediate-layer prototypes vs final-layer logit (ablation)Concept loss · ColorMNISTaccuracy (%): 25.55 -> 41.83Improved
use varying prototypes instead of fixed prototypes (ablation)Concept loss · ColorMNISTaccuracy (%): 40.02 -> 41.83Improved
increase teacher training data bias from 5% to 100% (ablation)Concept distillation · ColorMNISTstudent accuracy (%): 40.47 -> 23.63Hurt
100% biased teacher vs no distillation (ablation)Concept distillation · ColorMNISTstudent accuracy (%): 26.97 -> 23.63Hurt
use DINO ViT-B/8 teacher instead of ViT-S/8 (ablation)Concept distillation · ColorMNISTaccuracy (%): 39 -> 50.93Improved
train concept loss vs base model on bias concept TCAVOurs · BFFHQTCAV score of bias concept: 0.78 -> 0.13Improved
train concept loss on ColorMNISTOurs · ColorMNISTTCAV score of bias concept: 0.52 -> 0.21Improved

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