| CARVIEW |
Select Language
HTTP/2 200
server: GitHub.com
content-type: text/html; charset=utf-8
last-modified: Tue, 05 Mar 2024 07:57:38 GMT
access-control-allow-origin: *
etag: W/"65e6d072-23de"
expires: Tue, 30 Dec 2025 09:57:21 GMT
cache-control: max-age=600
content-encoding: gzip
x-proxy-cache: MISS
x-github-request-id: 1878:292AC1:9E7E74:B20077:69539FA7
accept-ranges: bytes
age: 0
date: Tue, 30 Dec 2025 09:47:21 GMT
via: 1.1 varnish
x-served-by: cache-bom-vanm7210030-BOM
x-cache: MISS
x-cache-hits: 0
x-timer: S1767088041.916850,VS0,VE209
vary: Accept-Encoding
x-fastly-request-id: 205f8fadc8100884832e33271c4c2a15a368b28f
content-length: 3039
Welcome to Deli Zhao's Homepage
 |
Deli Zhao | 赵德丽 |
About Me
-
Deli Zhao is leading the AI team in Alibaba DAMO Academy. Working with Dr. Jingren Zhou, my team has released Alibaba Tongyi big models such as Composer 2.0, VideoComposer 1.0, and I2VGen-XL for customizable and creative image generation and video synthesis, respectively. Before joining Alibaba, I worked at HTC and Xiaomi as the research director for six years. I also spent six wonderful years doing research in the Visual Computing Group of Microsoft Research Asia (MSRA) and in the Multimedia Lab (MMLab) at the Chinese University of Hong Kong (CUHK), working with Prof. Xiaoou Tang. I have performed research on computer vision and machine learning for nearly two decades, now mainly focusing on generative models, multi-modal learning, and foundation models. I am also interested in the fundamental principle of deep learning (two theoretical papers coworking with Prof. Michael Jordan). I closely work with Dr. Yujun Shen on generative models.
News
- [11/2023]: We released our new video generation model I2VGen-XL!
- [09/2023]: 5 papers accepted by NeurIPS. The acceptance rate in my team is 55.6% (official 26.1%).
- [09/2023]: Keynote on "Towards On-Chain AI" at Shanghai Blockchain Week. The slides are available here.
- [07/2023]: 12 papers accepted by ICCV. The acceptance rate in my team is 70.6% (official 26.8%).
- [06/2023]: We released VideoComposer, the Alibaba big model on video generation and editing.
- [06/2023]: Keynote and round-table discussion on AIGC and design at U Design Week.
- [04/2023]: 3 papers accepted by ICML. The acceptance rate in my team is 75% (official 27.9%).
- [04/2023]: Talk on Composer at The University of Hong Kong.
- [04/2023]: Talk on Composer at The Hong Kong University of Science and Technology.
- [04/2023]: Keynote on AIGC at Hong Kong Web3 Festival.
- [02/2023]: 10 papers accepted by CVPR.
- [02/2023]: We released Composer, the Alibaba big model on image generation and editing.
Research
Generative Foundation Models
- Text-to-image: Composer, Cones, AnyDoor
- Text-to-video: VideoComposer, VideoFusion, FaceComposer
- Image-to-video: I2VGen-XL
- Diffusion model: Dimensionality-Varying Diffusion Process, Eliminating Lipschitz Singularities
- Generative Adversarial Network: Latently Invertible Autoencoder (LIA), IDInvert, LowRankGAN, ReSeFa, LinkGAN
Multi-Modal Learning
Deep-Learning Theory
- Deep neural networks: Rank Diminishing in DNNs (cowork with Michael Jordan), Neural Dependencies (cowork with Michael Jordan)
- Generative models: Noise injection in GANs, Uncertainty Principles of Encoding GANs
Projects on AIGC
- Text-to-image: Alibaba Tongyi Wanxiang (通义万相)
- Text-to-video: Video generation, Open-source models on Hugging Face, Open-source models on ModelScope
My Book (in writing)
 |
|