TY  - RPRT
TI  - EVA: Bridging Performance and Human Alignment in Hard-Attention Vision Models for Image Classification
AU  - Pengcheng Pan
AU  - Yonekura Shogo
AU  - Kuniyoshi Yasuo
PY  - 2026
UR  - https://arxiv.org/abs/2603.27340
ID  - 2603.27340
ER  -