Yi-Fu Wu

I am a PhD candidate at Rutgers University advised by Professor Sungjin Ahn. I was previously a Student Researcher at Google and have interned at Generally Intelligent (now known as Imbue) and Apple. Before starting my PhD, I was a Founding Engineer at Boxed and led the Platform Engineering team. I received an MS from Stanford University and a BS from the University of Illinois, Urbana-Champaign.

My recent research has been focused on learning structured representations of the world from images and video.

email: yifu.[lastname] AT {rutgers, gmail}

Publications

• Neural Language of Thought Models [paper]
  Yi-Fu Wu, Minseung Lee, Sungjin Ahn
  ICLR 2024
  Also appeared as "Object-Centric Semantic Vector Quantization" in: Causal Representation Learning Workshop and Unifying Representations in Neural Models Workshop at NeurIPS 2023
• Inverted-Attention Transformers Can Learn Object Representations: Insights From Slot Attention [paper]
  Yi-Fu Wu, Klaus Greff, Gamaleldin F. Elsayed, Michael C. Mozer, Thomas Kipf, Sjoerd van Steenkiste
  Causal Representation Learning Workshop and Unifying Representations in Neural Models Workshop at NeurIPS 2023
• An Investigation into Pre-Training Object-Centric Representations for Reinforcement Learning [website] [paper]
  Jaesik Yoon, Yi-Fu Wu, Heechul Bae, Sungjin Ahn
  ICML 2023
• Simple Unsupervised Object-Centric Learning for Complex and Naturalistic Videos [website] [paper]
  Gautam Singh, Yi-Fu Wu, Sungjin Ahn
  NeurIPS 2022
• TransDreamer: Reinforcement Learning with Transformer World Models [paper]
  Chang Chen, Jaesik Yoon, Yi-Fu Wu, Sungjin Ahn
  Deep RL Workshop at NeurIPS 2021
• Generative Video Transformer: Can Objects be the Words? [paper]
  Yi-Fu Wu, Jaesik Yoon, Sungjin Ahn
  ICML 2021
• Improving Generative Imagination in Object-Centric World Models [website] [paper] [code]
  Zhixuan Lin, Yi-Fu Wu, Skand, Bofeng Fu, Jindong Jiang, Sungjin Ahn
  ICML 2020
• SPACE: Unsupervised Object-Oriented Scene Representation via Spatial Attention and Decomposition [website] [paper] [code]
  {Zhixuan Lin, Yi-Fu Wu, Skand, Weihao Sun}, Gautam Singh, Fei Deng, Jindong Jiang, Sungjin Ahn
  ICLR 2020