Feng (Jeff) Liang, 梁丰

I am a Research Scientist at Meta AI. I obtained my PhD from UT Austin with Prof. Diana Marculescu. Prior to that, I completed my master's at Tsinghua University and my bachelor's at Huazhong University of Science and Technology.

I’m interested in building efficient, personalized, and creative AI for everyone. My key research directions include:

• Efficient AI: Scalable learning/inference under data, compute, and energy constraints.
• Personalized AI: Developing privacy-preserving, user- and context-adaptive models.
• AI for Creativity: Empowering human creativity via controllable multimodal generation and understanding.

Email  /  CV  /  Google Scholar  /  Linkedin  /  Zhihu  /  Twitter

We present Movie Weaver to support multi-concept video personalization.

We present StreamV2V to support real-time video-to-video translation for streaming input.

We leverage the temporal optical flow clue within video to enhance the temporal consistency for text guided video-to-video synthesis.

For the first time, we show open-vocabulary generalist models match the performance of supervised specialist models without dataset-specific adaptations.

SupMAE extends MAE to a fully-supervised setting by adding a supervised classification branch, thereby enabling MAE to effectively learn global features from golden labels.

We propose Data efficient CLIP (DeCLIP), a method to efficiently train CLIP via utilizing the widespread supervision among the image-text data.