精品英语文档：指导老师.ppt

AI at IBM Research 背 景 50年以前，IBM就已經開始研究AI了 今日，IBM已在這領域擁有領導者的地位 結合科技的應用，去解決各種困難的問題 起 源 1950年代 Arthur Samuels就發現，電腦程式化的西洋棋，足夠打敗人腦 同時，Herbert Gelernter也發現，利用圖表資訊和幾何學公式解決一個複雜幾何問題 起 源 AI研究是在1970 1980年代，突飛猛進 過去常用 symbolic paradigm; 今日常用 statistics statistics : machine learning and natural language processing (NLP) 衍伸性研究statistical learning theory and Bayesian network AI可以解決的問題 Computer algebra systems Robotics Conceptual graphs Knowledge representation (specifically K-REP) Expert systems Epistemic logics Nonmonotonic reasoning IBM在AI領域專長 representation and reasoning, statistical AI vision, and game playing Representation and reasoning logic programming nonmonotonic reasoning Planning and scheduling intelligent agents. courteous logic programs courteous logic programs: let users specify the scope of potential conflicts by pairwise mutual exclusion. let them specify partially ordered priorities between rules, which can resolve conflicts between rules. FANs Formula-Augmented Networks :a knowledge structure that enables efficient reasoning about potentially conflicting business rules. extend semantic networks inheritance with exceptions by allowing the attachment of logical formulas at network nodes providing methods to determine which formulas “apply” to a particular node FANs In conflict，determine which to save and which to discard. The decision involves information inherent in the network’s topology FANs works well for businessdomains BenInq an expert system for benefits inquiry in the medical insurance industry a system to dynamically configure banking and insurance products Natural language in Zurich, Nobel laureate Gerd Binning and Juerg Klenk lead MeanX project anaphora resolution word disambiguation object identification MeanX input network fractal networks Knowledge base Agents cognitive psychology Output network Intelligent agents. they focuses on : development of core technology to quick implementation of