¡¾µçÐÅѧԺ¡¿Ñ§Êõ½²×ù£º Xue Li¡¶High-Order Reasoning with Large Language Model in Life-Critical Decisions¡·

±¨ ¸æ ÈË

Xue Li£¨Professor, The University of Queensland£©

±¨¸æʱ¼ä

2024Äê11ÔÂ27ÈÕ£¨ÖÜÈý£©ÍíÉÏ18:30

±¨¸æµØµã

½·½­Ð£Çø¿Æ¼¼×ÛºÏÂ¥401»áÒéÊÒ

±¨

¸æ

ÄÚ

ÈÝ

¼ò

½é

This talk is to report our recent research effort on how to use large language models for the high-order reasoning to help professionals to make timely decisions that are often life-critical.  

LLMs can be used to analyse complex contextual information, generate possible solutions, and provide recommendations or insights to humans making critical choices.

In low-order reasoning, search-based approaches draw conclusions on what-where-when-how-why type questions finding readily available answers in the solution space. In high-order reasoning, a conclusion is not simply to be derived or found but to be assessed based on the heuristics of a certain circumstance, where some preferences are considered, and outcomes are verified. In high-order reasoning, a sequence of conclusions is made available one-after-another and assessed consecutively. Thus, different algorithms or models can be applied, and different ranking criteria can be used to compare conclusions that can be exploited from the given evidence.

³Ð°ìѧԺ

µç×ÓÓëÐÅÏ¢¹¤³ÌѧԺ

·¢²¼ÈÕÆÚ

2024Äê11ÔÂ26ÈÕ

»¶Ó­¹ã´ó½Ìʦ¡¢Ñ§Éú²Î¼Ó£¡