AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

OCNR: Stabilizing Self-Play by Mitigating Iteration-Collapse With One-Class Novelty Rewards

Seungyoo Lee, Giung Nam, Hyungi Lee, Juho Lee

ICML 2026 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervised speech and brain decoding · 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
add one-class novelty reward (OCNR) to R-Zero questionerR-Zero · Math AVG, Iter 5average accuracy: 45.27 -> 52.22Improved
add OCNR to R-Zero for general-purpose reasoning transferR-Zero · MMLU-Proaccuracy: 54.7 -> 60.2Improved
add OCNR to Absolute Zero proposerAZR · AMCaccuracy: 37.5 -> 42.5Improved
add OCNR to Absolute Zero proposerAZR · HumanEval+pass@1: 71.3 -> 72.5Improved
double the over-generated data used for solver training (R-Zero 2x) without novelty reward (ablation)R-Zero · Math AVG, Iter 5average accuracy: 45.27 -> 44.02Hurt
double data for R-Zero (2x) at early iteration (ablation)R-Zero · Math AVG, Iter 1average accuracy: 46.50 -> 47.52Improved
use sigmoid-based novelty reward instead of distance-based (ablation)OCNR · t-SNE of generated questionsquestion noveltyHurt
use more prototypes in the Seen Detector (K=1 to K=3) (ablation)OCNR Seen Detector · seen detectionAUROC: 0.9152 -> 0.9321No clear effect
add novelty coefficient lambda_nov >= 0.1 (ablation)R-Zero · Math AVG, Iter 2 to Iter 5average accuracy: 48.10 -> 45.27 for lambda 0.0Improved
extend R-Zero with OCNR to 10 iterationsR-Zero w/ OCNR · Math AVG, Iter 7 to Iter 10average accuracy: 54.51 -> 53.98Improved

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