A Survey on User-Centric Personalization in LLMs

PhD Qualifying Examination


Title: "A Survey on User-Centric Personalization in LLMs"

by

Miss Yeongseo JUNG


Abstract:

The remarkable advancements in large language models (LLMs) have 
revolutionized natural language processing across diverse tasks. As these 
models continue to evolve, expectations have shifted from their 
general-purpose functionality to tailoring LLM capabilities for individual 
users, enabling personalized and context-aware interactions. However, 
personalizing LLMs to meet individual user preferences remains a significant 
challenge. Designed to generalize over diverse datasets, LLMs often lack 
mechanisms to effectively integrate fine-grained, user-specific knowledge. 
Moreover, despite progress in computational efficiency, adapting these models 
for unique user needs is constrained by the resource-intensive nature of 
user-specific training and limitations in context handling.

This survey explores these challenges and examines recent advancements in LLM 
personalization. We present strategies for modeling multiple users and 
integrating them into foundational models to enhance personalization while 
addressing key issues of scalability, efficiency, and adaptability.


Date:                   Wednesday, 18 December 2024

Time:                   11:00am - 12:00noon

Venue:                  Room 2129B
                        Lift 19

Committee Members:      Prof. Lei Chen (Supervisor)
                        Prof. Ke Yi (Chairperson)
                        Prof. Qiong Luo
                        Dr. Huiru Xiao