SS06: Coding Anything for AI Tasks
Motivation & significance
•Coding Anything for AI Tasks has emerged as a key area of research in multimedia. This concept expands beyond visual perception to include LLM-related representations, reflecting the rapid development of LLMs.
•Coding for AI tasks prioritizes optimizing coding efficiency while maintaining AI task performance, which is essential for AI-driven systems.
Topics of interest
•Visual coding for single or multiple AI tasks
•Visual coding for both AI tasks and human visual perception
•Feature coding for single or multiple AI tasks
•LLM-related representation (such as prompt, token, etc.) compression
•Semantic distortion measurement methods in coding for AI applications Novel frameworks, benchmarks, and datasets for coding for AI applications
Organizers
Changsheng Gao, Research Fellow, Nanyang Technological University, Singapore, email: changsheng.gao@ntu.edu.sg
Xin Jin, Assistant Professor, Eastern Institute of Technology, China, email: jinxin@eitech.edu.cn
Jian Jin, Research Fellow, Nanyang Technological University, Singapore, email:jian.jin@ntu.edu.sg
Nam Ling, Professor, Santa Clara University, USA, email: nling@scu.edu