摘要: We report on initial experiments with an implemented learning system whose inputs are images of two-dimensional shapes. The first builds semantic network shape descriptions based Brady's smoothed local symmetry representation. It learns models from them using a modified version Winston's ANALOGY program. program uses only positive examples, and is capable disjunctive concepts. discuss the lcarnability descriptions.
ijcai.org参考文章(27)
Michael Brady, Philip E Agre, None, The mechanic's mate european conference on artificial intelligence. pp. 79- 94 ,(1984)
Patrick H. Winston, Learning Structural Descriptions From Examples The Psychology of Computer Vision. ,(1970)
Patrick H Winston, Learning by Augmenting Rules and Accumulating Censors. ,(1982)
Thomas G. Dietterich, Ryszard S. Michalski, Inductive learning of structural descriptions: Evaluation criteria and comparative review of selected methods Artificial Intelligence. ,vol. 16, pp. 257- 294 ,(1981) , 10.1016/0004-3702(81)90002-3
Margaret Morrison Fleck, Local Rotational Symmetries IEEE. pp. 332- 337 ,(1985)
Michael Brady, Criteria for Representations of Shape Human and Machine Vision. pp. 39- 84 ,(1983) , 10.1016/B978-0-12-084320-6.50008-6
M Halle, The Sound Pattern of English ,(1968)
Kurt Alan Vanlehn, Felicity conditions for human skill acquisition: validating an ai-based theory Massachusetts Institute of Technology. ,(1983)
John Canny, Finding Edges and Lines in Images Masters Thesis. pp. 16- ,(1983)
Brachman, Fikes, Levesque, Krypton: A Functional Approach to Knowledge Representation IEEE Computer. ,vol. 16, pp. 67- 73 ,(1983) , 10.1109/MC.1983.1654200