FOMO-VL 2022 : The 1st Workshop on Foundation Models for Vision and Language

The FOMO-VL 2022 workshop aims to bring together practitioners and researchers with a specific focus on the emerging trends and industry needs associated with multimodality data analytics with foundation models. Both theoretical and experimental submissions are encouraged. Papers should elaborate on model pre-training and adaptation methods with multimodality data, opportunities and issues associated with foundation models, visualization and efficient large-scale training tools, methods, and novel applications or systems. Topics of interest include but are not limited to:

1. Theories and algorithms of self-supervised learning, e.g., generative and contrastive approaches
2. Scaling and generalization of pre-training including multi-task and modularized architectures
3. Efficient distributed training technique for big multimodality data
4. Light-weight model adaption on resource-limited devices and scenarios
5. Data-efficient model adaptation methods: zero-shot and few-shot
6. Vision-and-language (V+L) benchmarks and evaluation
7. Knowledge-enriched methods
8. Interactive AI agents with foundation models
9. Foundation models beyond V+L, e.g., structured data, multilingual, video and knowledge-graph
10. Data collection for foundation models
11. Risks and bias issues in foundation models
12. Novel applications in domains including retails, finance, and healthcare
13. Visions/Comments on the futures of foundation models for V+L

Submission Guidelines We welcome full research papers (be limited to a maximum of 8 pages excluding supplementary materials), as well as vision/demo/poster/industrial papers (up to 3 pages excluding references and appendix). Submissions longer than 8 main pages will be rejected without review. You can include any number of pages for references and appendix. If you have an appendix, please combine it with the main pages into a single PDF file, as no additional file will be accepted in the submission system. All submissions will be reviewed by the Program Committee on the basis of technical quality, relevance to scope of the conference, originality, significance, and clarity.

Panelists (random order):
-- Jianfeng Gao (MSR)
-- Trishul Chilimbi (Amazon)
-- Christoph Schuhmann (LAION)
-- Ruslan Salakhutdinov (CMU)
-- Ludwig Schmidt (UW)

Invited Speakers (random order):
-- Danqi Chen (Princeton)
-- Xifeng Yan (UCSB)
-- Tengyu Ma (Standford)
-- Letitia Parcalabescu (University of Heidelberg)
-- Jiahui Yu (Google)
-- Lu Yuan (MSR)
-- Jiasen Lu (Allen Institute of AI)
-- Justin Lin (Alibaba)