Accès ouvert
2026
preprint
OpenAlex
Renshu Gu, Jialiang Chen, Fei Gao, Hang Su et autres
Weakly supervised segmentation relies heavily on class activation maps (CAMs) to initially localize target regions. However, CAMs are often noisy and prone to catastrophic failures. Existing remedies typically introduce additional training stages or prototype learning, increasing computational cost and reducing robustness. In …
Accès ouvert
2026
preprint
OpenAlex
Hao Cheng, Hang Su, Y Yao
We propose a general necessary condition for a spinful fermion chain with SU(2) spin-rotation symmetry to be gapped. Specifically, we prove that the expectation value of a properly defined fermionic twisting operator asymptotically approaches unity in any gapped phase with finite ground-state …
Accès ouvert
2026
preprint
OpenAlex
Guangyan Chen, Meiling Wang, Te Cui, Zichen Zhou et autres
The ability to acquire skills rapidly and effortlessly while retaining those already mastered is essential for robots. However, current methods still rely on a cumbersome training-time loop that is costly and slow, while eroding skills already mastered. In this paper, we introduce …