Back to reviews
min readarXiv:2603.04731v1

When Priors Backfire: On the Vulnerability of Unlearnable Examples to Pretraining

Authors: Zhihao Li, Gezheng Xu, Jiale Cai, Ruiyi Fang, Di Wu

Pending (κ=0.55)Intermediaterepresentationspuriousalignment

RSCT Score Breakdown

Relevance (R)
0.38
Superfluous (S)
0.32
Noise (N)
0.31

TL;DR

When Priors Backfire: On the Vulnerability of Unlearnable Examples to Pretraining...

When Priors Backfire: On the Vulnerability of Unlearnable Examples to Pretraining

RSCT Certification: κ=0.550 (certified) | RSN: 0.38/0.32/0.31 | Topics: representation, spurious, alignment

Overview

This paper addresses topics relevant to RSCT research, specifically in the areas of representation, spurious, alignment.

Key RSCT Relevance:

  • Topic similarity score: 40%
  • RSCT whitepaper similarity: 32%
  • Combined relevance: 35%

RSCT Quality Metrics

| Metric | Value | Interpretation | |--------|-------|----------------| | κ-gate | 0.550 | Certified | | R (Relevance) | 0.375 | Direct relevance to research goals | | S (Stability) | 0.318 | Supporting context and patterns | | N (Noise) | 0.307 | Irrelevant components | | Gate Reached | 4 | Certification depth |

Paper Details

  • Authors: Zhihao Li, Gezheng Xu, Jiale Cai, Ruiyi Fang, Di Wu
  • Source: arxiv
  • Primary Topic: representation
  • Difficulty: Intermediate

This review was auto-generated by the Swarm-It RSCT discovery pipeline.

About This Review

This review was auto-generated by the Swarm-It research discovery platform. Quality is certified using RSCT (RSN Certificate Technology) with a κ-gate score of 0.55. RSN scores: Relevance=0.38, Superfluous=0.32, Noise=0.31.