Adapting large language models to individual users remains challenging due to the tension between fine-grained personalization and scalable deployment. We present CARD, a hierarchical framework that achieves effective personalization through progressive refinement. CARD first clusters users according to shared stylistic patterns and learns group-specific LoRA adapters, enabling robust generalization and strong low-resource performance. To capture individual differences within each cluster, we propose an implicit preference learning mechanism that contrasts user-authored text with cluster-level generations, allowing the model to infer user-specific style preferences without manual annotation. At inference time, CARD injects personalization exclusively at decoding via lightweight user preference vectors and low-rank logit corrections, while keeping the base model frozen. Experiments on the LaMP and LongLaMP benchmarks show that CARD achieves superior generation quality compared to baselines, while significantly improving efficiency and scalability for practical personalized text generation.
The Reward Model Selection Crisis in Personalized Alignment
School of Informatics, University of Edinburgh(信息学院,爱丁堡大学);Agency for Science, Technology and Research (A*STAR)(科技研究局);Samsung AI Research Center(三星人工智能研究中心)
Auditing and Mitigating Privacy Leakage in Cloud-Edge Collaborative Decoding
The Hong Kong Polytechnic University(香港理工大学);Institute for AI Industry Research, Tsinghua University(清华大学人工智能产业研究院);School of Software, Tsinghua University(清华大学软件学院)
Enhancing Commentary Strategies for Imperfect Information Card Games: A Study of Large Language Models in Guandan Commentary
Guangdong University of Technology(广东工业大学);East China Jiaotong University(华东交通大学);Emory University(埃默里大学);University of Minnesota - Twin Cities(明尼苏达大学双城分校);South China University of Technology(华南理工大学);University of Electronic Science and Technology of China(电子科技大学);Amazon(亚马逊公司);University of Toronto(多伦多大学)
Toward Practical BCI: A Real-time Wireless Imagined Speech EEG Decoding System
Dept. of Artificial Intelligence Korea University Seoul, Republic of Korea jiha\;Dept. of Artificial Intelligence Korea University Seoul, Republic of Korea hg\;Cognitive Engineering Korea University Seoul, Republic of Korea gh\;Tech. Innovation Group KT R\&D Center Seoul, Republic of Korea;Dept. of Artificial Intelligence Korea University Seoul, Republic of Korea
One Chatbot Per Person: Creating Personalized Chatbots based on Implicit User Profiles
Renmin University of China(中国人民大学);Beijing Key Laboratory of Big Data Management and Analysis Methods(大数据管理与分析方法北京市重点实验室);Key Laboratory of Data Engineering and Knowledge Engineering, MOE(教育部数据工程与知识工程重点实验室);Université de Montréal(蒙特利尔大学)
Origin-Aware Next Destination Recommendation with Personalized Preference Attention
GrabTaxi Holdings(GrabTaxi控股公司);School of Computing, National University of Singapore(新加坡国立大学计算学院);National University of Singapore(新加坡国立大学);Grab-NUS AI Lab(Grab-新加坡国立大学人工智能实验室)
ECGTwin: Personalized ECG Generation Using Controllable Diffusion Model
State Key Laboratory of General Artificial Intelligence, Beijing, China(一般人工智能国家重点实验室);School of Intelligence Science and Technology, Peking University, Beijing, China(智能科学与技术学院);National Institute of Health Data Science, Peking University, Beijing, China(健康数据科学国家研究院);Department of Cardiology, Peking University People’s Hospital, Beijing, China(心血管内科)