AI Paper Atlas

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

Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanisms

Hyunjin Cho, Youngji Roh, Jaehyung Kim

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 0 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
cluster jointly in semantic and attribution views (RD) instead of K-means in one viewRD-clustering · AmbigQA, Qwen3-8Bmixed distortion Dγ: lowest in 94% of settingsImproved
use semantic-only K-means (KM-S)K-means · AmbigQA, Qwen3-8Bmechanistic distortionHurt
use attribution-only K-means (KM-A)K-means · AmbigQA, Qwen3-8Bsemantic coherenceHurt
steer with RD cluster medoid direction instead of semantic K-means medoidcluster steering · AmbigQA, Qwen3-8BSpearman correlation of steering: -0.02 -> 0.30Improved
steer with RD medoid instead of a random single continuation's attributioncluster steering · AmbigQA, Qwen3-8BSpearman correlation of steering: 0.13 -> 0.30Improved
use RD medoid instead of single continuation on held-out production-funding clustercluster steering · Table 3 case studyheld-out centered logit change (amplification): +0.068 -> +0.598Improved
increase beta from 0.5 to 0.75 to 1.0RD-clustering · 'If God was one of us' promptnumber of clusters: K=3 -> K=5 -> K=9Mixed
lower gamma to 0.1 (weight attribution view)RD-clustering · AmbigQA qualitativecluster structureMixed
raise gamma to 0.9 (weight semantic view)RD-clustering · AmbigQA qualitativenumber of clustersMixed
add Split operationRD-clustering · AmbigQA examplerate-distortion objective L_RD: 4.61 -> 4.57Improved
add Junk operationRD-clustering · AmbigQA examplerate-distortion objective L_RD: 5.54 -> 5.42Improved

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