Qualifying Exam
Qualifying ExamIntegrate Logical Reasoning and Machine Learning for Decision Making |
|
||
Friday, April 14, 2023, 10:00am - 12:00pm |
|||
Abstract:
The ability to engage in logical reasoning is a crucial aspect of human intelligence, which also plays a significant role in both traditional machine learning models and large language models (LLMs). However, the relationship between logical reasoning and these models has not been extensively explored. In this presentation, we will explore the potential applications of logical reasoning in both traditional models and LLMs.
In the case of traditional models, by integrating two important types of reasoning ability---counterfactual reasoning and logical reasoning---we propose Counterfactual Collaborative Reasoning (CCR), which conducts counterfactual logic reasoning to improve the performance. In particular, we use the recommender system as an example to show how CCR alleviates data scarcity, improves accuracy and enhances transparency.
For LLM, we design a Logical Large Language Model (L3M) that integrates the strengths of logical reasoning and large language models. The data in L3M is represented in logical expressions and the model uses logical constraints to learn the rules of basic logical operations such as And, Or, and Not. We conduct experiments on both theoretical tasks (solving logical equations) and practical tasks (recommender systems). The results of our theoretical experiments demonstrate that L3M is highly effective in solving logical expressions and variables.
Speaker: Jianchao Ji
Location : CoRE 305
Committee:
Professor Yongfeng Zhang (Advisor)
Professor Hao Wang
Professor Dong Deng
Professor Sudarsan Kannan
Event Type: Qualifying Exam
Abstract: See above
Organization:
Rutgers University
School of Arts & Sciences
Department of Computer Science
Contact Professor Yongfeng Zhang