AI RESEARCH
Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift
arXiv CS.CV
•
ArXi:2512.15647v3 Announce Type: replace Soft labels from teacher models are a de facto practice for knowledge transfer and large-scale dataset distillation (e.g., SRe2L, LPLD). However, when we limit the number of crops per image to reduce the substantial cost of storing precomputed soft labels, these methods suffer severely from local semantic drift: visually ambiguous crops can cause soft supervision to deviate from the image-level ground-truth semantics, leading to persistent errors and a train-test distribution mismatch.