Richard Cox [1] set the axiomatic foundations of probable inference and the algebra of propositions. He showed that consistency within these axioms requires certain rules for updating belief. In this paper we use the analogy between probability and utility introduced in [2] to propose an axiomatic foundation for utility inference and the algebra of preferences. We show that consistency within these axioms requires certain rules for updating preference. We discuss a class of utility functions that stems from the axioms of utility inference and show that this class is the basic building block for any general multiattribute utility function. We use this class of utility functions together with the algebra of preferences to construct utility functions represented by logical operations on the attributes.
PromptGraph: Graph-Guided Prompt Sanitization for Balancing Privacy and Utility in LLM Inference
School of Computer Science and Information Engineering, Hefei University of Technology（计算机科学与信息工程学院，合肥工业大学）；International College Beijing, China Agricultural University（北京国际学院，中国农业大学）
Defending Wearable VLMs Against Private Attribute Inference
Swanson School of Engineering, University of Pittsburgh（匹兹堡大学斯旺森工程学院）；North Allegheny Senior High（北阿勒格尼高级中学）；College of Engineering, Michigan State University（密歇根州立大学工程学院）；University of Central Florida（中佛罗里达大学）
SharedRequest: Privacy-Preserving Model-Agnostic Inference for Large Language Models
National University of Singapore (Chongqing) Research Institute（新加坡国立大学（重庆）研究院）；Chongqing Key Laboratory of Trusted Perception and Interaction Technology for Intelligent and Connected Vehicles（重庆智能网联车辆可信感知与交互技术重点实验室）；National University of Singapore（新加坡国立大学）；Hebrew University of Jerusalem（耶路撒冷希伯来大学）；State Key Laboratory of Intelligent Vehicle Safety Technology, Chongqing, China（重庆智能车辆安全技术国家重点实验室）；CHONGQING CHANGAN AUTOMOBILE Co., Ltd（重庆长安汽车有限公司）
