2005.5.25-27：美国天普大学计算机与信息科学系副教授王培教授访问我中心，作了人工智能与词项逻辑的报告，并和中心成员就人工智能中的推理问题进行了讨论。

Talk Abstract, Logic Seminar, Peking University

May 26, 2005

Intelligence, Logic, and Evidence

Pei Wang
Department of Computer and Information Sciences, Temple University
http://www.cis.temple.edu/~pwang/pei.wang@temple.edu

Motivation: Artificial Intelligence

The overall goal of AI is to find the mechanism of intelligence (thinking, cognition).

NARS: a general-purpose AI system that integrates various types of reasoning and learning.

Reference: Toward a unified artificial intelligence, AAAI Fall Symposium on Achieving Human-Level Intelligence through Integrated Research and Systems , Washington DC, October 2004.

Problem: Logic for empirical reasoning

The traditional mathematical logic runs into problems in AI.

Reference: Cognitive logic versus mathematical logic, Third International Seminar on Logic and Cognition, Guangzhou , May 2004.

Several wide-spread misconceptions of AI come from the confusion between the two types of reasoning systems.

Reference: Three Fundamental Misconceptions of Artificial Intelligence (draft).

Key issue: Evidence formalization

There are several ways to formalize the notion of "evidence", based on different assumptions:

·  Evidence in binary logic

·  Evidence and probability

·  Evidence and imprecise probability

·  Evidence in NARS

Reference: Formalizations of Evidence (draft).

Solution: Non-Axiomatic Reasoning System

Theoretic summary: manuscript

Implementation in Java and Prolog: demonstrations