网络时代人工智能研究与发展.pdfVIP

4 1 V o.l 4 . 1 2009 2 CAA I T ransactions on Intelligent System s F eb. 2009 李德毅 (总参第 61研究所, 北京 100840 : 50 , , . , , , ; , , ; , , , . 3, . : ; ; ; : TP 18 : A : 167 347 85( 2009 A I research and developm ent in the network age L I Deyi ( Ch ina Institute o f E lectron ic System Eng inee ring, Be ij ing 100840, Ch ina Abs tract: Pattern recogn it ion, know ledge eng ineering, and robotics have m ade significant progress in the 50year histo ry of artificial in telligence, yet A I disp lays far from hum an intelligence. In the current netw ork era, if re searchers in artif icial intelligence w ant to m ax mi ize developm ents and innovat ions in interd isc ip linary studies, they m u st pay m ore attention to the intersect ions and infiltrat ions of cognitive science, brain science, physics, network science, compu ter science, and artificial inte lligence. Research in cognitive physics w ill be an especially mi portant direction in A I. N atura l language is an mi portant objective in A I research; w e need to establish an uncertainty transform ation m odel that can quantitative ly represent its concepts. Th is d ictates that an A I sc ience w ith uncertainty w ill be deve loped. Considering the sm allw orld m odel and sca lefree features of com plex netw orks in real life, w e need to use ne wt ork topo logy as a new w ay for know ledge representation. T his w ill a id study o f the progress o f net w ork topology, newt ork dynam ics and inte lligence. T his paper d iscu sses these three d irections in detail. Som e con crete suggestions for further research are also prov ided. K ey