default author photo

Byeongho Heo

NAVER AI Lab

SCHOLARLY PAPERS

1

DOWNLOADS

229

TOTAL CITATIONS

0

Scholarly Papers (1)

1.

Improving ViT Interpretability with Patch-Level Mask Prediction

Number of pages: 7 Posted: 19 Apr 2023
Junyong Kang, Byeongho Heo and Junsuk Choe
Sogang University, NAVER AI Lab and Sogang University - Department of Computer Science and Engineering
Downloads 229 (333,232)

Abstract:

Loading...

Machine Learning, Deep Learning, vision transformer, Interpretability, Weak supervision, Object localization