Selected Publications
V-Express: Conditional Dropout for Progressive Training of Portrait Video Generation;
Cong Wang*,
Kuan Tian*,
Jun Zhang†,
Yonghang Guan,
Feng Luo,
Fei Shen,
Zhiwei Jiang†,
Qing Gu,
Xiao Han,
Wei Yang;
arXiv:2406.02511.
[code]
[project page]
[arXiv]
[models]
TL;DR: V-Express aims to generate a talking head video under the control of a reference image, an audio, and a sequence of V-Kps images.
Advanced Sign Language Video Generation with Compressed and Quantized Multi-Condition Tokenization;
Cong Wang*,
Zexuan Deng*,
Zhiwei Jiang†,
Fei Shen,
Yafeng Yin,
Shiwei Gan,
Zifeng Cheng,
Shiping Ge,
Qing Gu;
arXiv:2506.15980.
[code]
[arXiv]
TL;DR:
We propose SignViP, a novel SLVG framework that incorporates
multiple fine-grained conditions for improved generation fidelity, which adopts a
discrete tokenization paradigm to integrate and represent the conditions.
Ensembling Diffusion Models via Adaptive Feature Aggregation;
Cong Wang*,
Kuan Tian*,
Yonghang Guan,
Fei Shen,
Zhiwei Jiang†,
Qing Gu,
Jun Zhang†;
International Conference on Learning Representations (ICLR), 2025.
[paper]
[code]
[poster]
[arXiv]
TL;DR: We propose Adaptive Feature Aggregation (AFA) to ensemble multiple diffusion models dynamically based on different states like prompts, noises, and spatial locations.
Aggregating Multiple Heuristic Signals as Supervision for Unsupervised Automated Essay Scoring;
Cong Wang,
Zhiwei Jiang†,
Yafeng Yin,
Zifeng Cheng,
Shiping Ge,
Qing Gu;
Annual Meeting of the Association for Computational Linguistics (ACL), 2023.
[paper]
[code]
[poster]
[slides]
[video]
TL;DR: We propose ULRA for unsupervised automated essay scoring, which utilizes multiple heuristic quality signals to train a neural network using Deep Pairwise Rank Aggregation loss.
Controlling Class Layout for Deep Ordinal Classification via Constrained Proxies Learning;
Cong Wang,
Zhiwei Jiang†,
Yafeng Yin,
Zifeng Cheng,
Shiping Ge,
Qing Gu;
AAAI Conference on Artificial Intelligence (AAAI), 2023.
[paper]
[code]
[poster]
[slides]
[arXiv]
TL;DR: We propose Constrained Proxies Learning for deep ordinal classification, which learns proxies for ordinal classes and adjusts their layout in feature space to capture ordinal relationships.