Open access
Jul 2026
Test Case Prioritization for DNNs via Neural Collapse Instability
A Neural-Collapse-Inspired Prioritization (NCIP) framework that replaces absolute confidence with cross-checkpoint prediction variability in the terminal training regime, where model geometry becomes highly structured and achieves strong performance in early fault discovery compared with competitive baselines.
Chun-Yu Liu, Ming-Yuan Li, Yang Li et al.
· arXiv.org · 0 citations