Active data selection in supervised and unsupervised learning

作者: Martina Hasenjäger

DOI:

关键词:

摘要: In the context of computer science, learning is applied in situations that are so complex conventional programming techniques either unavailable or not practical. But often empirical knowledge on phenomenon/process under consideration available form data from repeated measurements. Learning then means extracting basic regularities these and thus can be seen as building an abstraction phenomenon yields a complete robust description its interesting aspects. The success process depends number factors such concrete system, procedure used for and, finally, at learner's disposal. It this last point selection main focus thesis. general, selected random way they capture aspects phenomenon. This necessarily most efficient acquisition, since does receive feedback learner. The may therefore tune with his current state it could be. In thesis, we discuss new paradigm aims improving efficiency neural network training procedures: active learning. Here, learner enabled to make use information already select those he expects informative. case, no longer passive recipient information, but takes role data. After review art learning, turn binary classification tasks. study detail approach problem based concepts theory. We develop heuristic algorithm local models, class learners up now has been considered context. Finally, extend area application unsupervised learning: propose topographic pairwise clustering founded statistical decision theory. Our results show computationally expensive that, comparison selection, strategies lead considerable reduction necessary samples. makes viable alternative, especially when cost acquisition high.

参考文章(132)
K. Lang, Learning to tell two spirals apart Proceedings of the 1988 Connectionist Models Summer School. ,(1989)
A. J. Kinderman, J. F. Monahan, New methods for generating student's t and gamma variables Computing. ,vol. 25, pp. 369- 377 ,(1980) , 10.1007/BF02285231
S. B. Thrun, The role of exploration in learning control Handbook of Intelligent Control: Neural, Fuzzy and Adaptive Approaches. ,(1992)
Teuvo Kohonen, Self-organized formation of topologically correct feature maps Biological Cybernetics. ,vol. 43, pp. 509- 521 ,(1988) , 10.1007/BF00337288
Ido Dagan, Sean P. Engelson, Selective Sampling In Natural Language Learning ,(1995)
Geoffrey W. Gates, The Reduced Nearest Neighbor Rule ,(1998)
Prasad Tadepalli, Ray Liere, Active Learning with Committees for Text Categorization national conference on artificial intelligence. pp. 591- 596 ,(1997)
J. Kangas, S. Kaski, T. Kohonen, Bibliography of Self-Organizing Map (SOM) Papers: 1981-1997 Neural Computing Surveys. ,vol. 1, pp. 1- 176 ,(1998)
Joachim M. Buhmann, Stochastic algorithms for exploratory data analysis: data clustering and data visualization Proceedings of the NATO Advanced Study Institute on Learning in graphical models. pp. 405- 419 ,(1998) , 10.1007/978-94-011-5014-9_14
S.R. Kulkarni, S.K. Mitter, J.N. Tsitsiklis, Active Learning Using Arbitrary Binary Valued Queries Machine Learning. ,vol. 11, pp. 23- 35 ,(1993) , 10.1023/A:1022627018023