Reproduce “Personalized Multimodal Large Language Models: A Survey” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Junda Wu, Hanjia Lyu, Yu Xia, Zhehao Zhang, Joe Barrow, Ishita Kumar, Mehrnoosh Mirtaheri, Hongjie Chen and 19 more

This survey paper provides a comprehensive review of personalized Multimodal Large Language Models (MLLMs), introducing a taxonomy that categorizes personalization techniques across four key application areas and mechanisms. It synthesizes existing research, identifies relevant datasets, and outlines critical open challenges to guide future work in the field.

Key insights

  • Personalization in MLLMs can be systematically achieved through distinct mechanisms like instruction, alignment, generation, and fine-tuning, applicable across diverse multimodal tasks.
  • Existing personalized MLLM research predominantly focuses on text and image modalities, indicating a need for greater diversity in supported modalities (e.g., audio, video, graphs).
  • Significant gaps remain in the availability of robust multimodal benchmark datasets with user-specific information and in developing direct evaluation metrics for generation quality.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.