UniDiff: Advancing Vision-Language Models with Generative and Discriminative Learning.
Xiao DongRunhui HuangXiaoyong WeiZequn JieJianxing YuJian YinXiaodan LiangPublished in: CoRR (2023)
Keyphrases
- discriminative learning
- language model
- generative model
- probabilistic model
- language modeling
- n gram
- information retrieval
- query expansion
- maximum margin
- retrieval model
- highly efficient
- computer vision
- object categories
- structured prediction
- graphical models
- weakly supervised
- pairwise constraints
- image set
- prior knowledge
- unsupervised learning
- latent dirichlet allocation