Xinyu Zhang

Dr. Xinyu Zhang is a Research Fellow at the Australian Institute for Machine Learning (AIML), University of Adelaide's Faculty of Sciences, Engineering and Technology. Her research bridges computer vision and machine learning, focusing on image/video generation, self-supervised learning, and multimodal retrieval for human-centric AI applications.

Zhang's current investigations include:

• Causal representation learning and multimodal integration
• Bayesian deep learning frameworks
• Video generation with temporal consistency
• Lightweight detection transformers
• Unsupervised person re-identification

Analysis of recent publications reveals strong emphases on generative modeling innovations (especially video synthesis), efficient transformer architectures for real-time applications, and self-supervised representation learning. Her work frequently addresses the alignment between latent representations and human perception across vision-language tasks.

Dr. Zhang co-supervises graduate students on projects involving multi-agent 3D scene generation and knowledge transfer in low-supervision learning. She serves as conference reviewer for premier venues including CVPR, ICCV, and NeurIPS, contributing to the advancement of computer vision research.