This work reformulate homogeneous distance minimization as a bi-level optimization problem via a matching-and-approximating paradigm and introduces a difficulty-aware marginal reweighting mechanism to emphasize informative instances while preserving global geometric consistency.
Xiao Cui, Yu-Lei Qin, Mo Zhu et al.· IEEE Transactions on Pattern...· 1 citation
Pixel-space diffusion models provide an appealing alternative to latent-space diffusion models by directly modeling raw images without relying on a learned tokenizer. However, scaling pixel-space models has predominantly followed dense strategies, tightly coupling model capacity with active computation. In this work, w...
Rui-Xiao Dong, Jian-Min Bao, Bo-Zhi Luan et al.· 0 citations
Progressive autoregressive image codecs provide an appealing paradigm for generative compression by quantizing continuous latents into discrete tokens, transmitting coarse-to-fine prefix tokens and generating the remaining suffix tokens at the decoder. However, their reconstruction quality is fundamentally limited by t...
Qing-Lei Yan, Rui-Xiao Dong, Yu-Tao Xie et al.· 0 citations
Groupwise Reward Geometry Conditioning (GRGC) is proposed, a two-stage framework that improves advantage construction by shaping student-side reward groups during both critic training and policy optimization, and introduces negligible overhead over GAD.
The Multi-Level Optimal Transport (MultiLevelOT) is introduced, a novel approach that advances the optimal transport for universal cross-tokenizer knowledge distillation and is robust to different student and teacher models across model families, architectures, and parameter sizes.
Xiao Cui, Mo Zhu, Yu-Lei Qin et al.· IEEE Transactions on Pattern...· 2 citations
This paper proposes a novel and principled multimodal fusion strategy for person Re-ID that skillfully and automatically fuses RGB features with gait silhouette features without the need for explicit spatial calibration and introduces a new synthetic gait person Re-ID dataset, SGPR.
Dong Xi, Li Li, Wen-Gang Zhou et al.· Multimedia Systems· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.