Publications (8)
Mirai: Autoregressive Visual Generation Needs Foresight
Yonghao Yu, Lang Huang, Zerun Wang +2
Autoregressive (AR) visual generators model images as sequences of discrete tokens and are trained with a next-token likelihood objective. This strict causal supervision optimizes…
Blow-ups of minimal surfaces in the Heisenberg group
Yonghao Yu
In this paper, we revise Monti's results on the blow-ups of H-perimeter minimizing sets in . Monti demonstrated that the Lipschitz approximation of the blow-up, after…
Stretchcam: Zooming Using Thin, Elastic Optics
Daniel C. Sims, Oliver Cossairt, Yonghao Yu +1
Stretchcam is a thin camera with a lens capable of zooming with small actuations. In our design, an elastic lens array is placed on top of a sparse, rigid array of pixels. This len…
Video-Mirai: Autoregressive Video Diffusion Models Need Foresight
Yonghao Yu, Lang Huang, Runyi Li +2
Causal video generators must predict from the past, but they need not learn only from it. In streaming autoregressive video diffusion, each emitted segment becomes a commitment tha…
Brain-Inspired Stepwise Patch Merging for Vision Transformers
Yonghao Yu, Dongcheng Zhao, Guobin Shen +2
The hierarchical architecture has become a mainstream design paradigm for Vision Transformers (ViTs), with Patch Merging serving as the pivotal component that transforms a columnar…
FedDetox: Robust Federated SLM Alignment via On-Device Data Sanitization
Shunan Zhu, Jiawei Chen, Yonghao Yu +1
As high quality public data becomes scarce, Federated Learning (FL) provides a vital pathway to leverage valuable private user data while preserving privacy. However, real-world cl…
BoostDream: Efficient Refining for High-Quality Text-to-3D Generation from Multi-View Diffusion
Yonghao Yu, Shunan Zhu, Huai Qin +1
Witnessing the evolution of text-to-image diffusion models, significant strides have been made in text-to-3D generation. Currently, two primary paradigms dominate the field of text…
Logit Poisoning Attack in Distillation-based Federated Learning and its Countermeasures
Yonghao Yu, Shunan Zhu, Jinglu Hu
Distillation-based federated learning has emerged as a promising collaborative learning approach, where clients share the output logit vectors of a public dataset rather than their…