Shuzhao Xie

I am a PhD student at Tsinghua SIGS, working with Prof. Zhi Wang on metaverse. I received my Master's degree in Computer Technology from Tsinghua University in 2023 and my Bachelor's degree in Computer Science and Technology from Beijing Normal University in 2020.

Email / CV [2023.9] / Google Scholar / Github / Calendar

Research Interests

My research is in the area of novel view synthesis, generative models, and serving system. I have experience in machine learning services, reinforcement learning, gaussian splatting, and generative models.

Publications

	MesonGS: Post-training Compression of 3D Gaussians via Efficient Attribute Transformation Shuzhao Xie, Weixiang Zhang, Chen Tang, Yunpeng Bai, Rongwei Lu, Shijia Ge, Zhi Wang European Conference on Computer Vision (ECCV), 2024 paper / project page / code
	SkyML: A MLaaS Federation Design for Multicloud-based Multimedia Analytics Shuzhao Xie, Yuan Xue, Yifei Zhu, Zhi Wang IEEE Transactions on Multimedia (TMM) , 2024 paper / code
	Expansive Supervision for Neural Radiance Field Weixiang Zhang, Shuzhao Xie, Shijia Ge, Wei Yao, Chen Tang, Zhi Wang arXiv, 2024 paper
	Tuning-Free Visual Customization via View Iterative Self-Attention Control Xiaojie Li, Chenghao Gu, Shuzhao Xie, Yunpeng Bai, Weixiang Zhang, Zhi Wang arXiv, 2024 paper
	RFQuant: Retraining-free Model Quantization via One-Shot Weight-Coupling Learning Chen Tang, Yuan Meng, Jiacheng Jiang, Shuzhao Xie, Rongwei Lu, Xinzhu Ma, Zhi Wang, Wenwu Zhu IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024 paper / supp / poster / code
	A Joint Approach to Local Updating and Gradient Compression for Efficient Asynchronous Federated Learning Jiajun Song, Jiajun Luo, Rongwei Lu, Shuzhao Xie, Bin Chen, Zhi Wang, Wenwu Zhu International European Conference on Parallel and Distributed Computing (Euro-Par), 2024 paper
	TextIR: A Simple Framework for Text-based Editable Image Restoration Yunpeng Bai, Cairong Wang, Shuzhao Xie, Chao Dong, Chun Yuan, Zhi Wang arXiv, 2023 paper In this work, we design an effective framework that allows the user to control the restoration process of degraded images with text descriptions. We use the text-image feature compatibility of the CLIP to alleviate the difficulty of fusing text and image features. Our framework can be used for various image restoration tasks, including image inpainting, image super-resolution, and image colorization.
	Cost Effective MLaaS Federation: A Combinatorial Reinforcement Learning Approach Shuzhao Xie, Yuan Xue, Yifei Zhu, Zhi Wang INFOCOM, 2022 paper / project page / code As naively using multiple ML APIs can lead to lower analytical accuracy and poor user experience, we employ reinforcement learning to select the optimal combination of MLaaSes, thus maximizing accuracy while minimizing cost.