论文信息 - Learning By Error-Driven Decomposition

Learning By Error-Driven Decomposition

In this paper we describe a new selforganizing decomposition technique for learning high-dimensional mappings. Problem decomposition is performed in an error-driven manner, such that the resulting subtasks (patches) are equally well approximated. Our method combines an unsupervised learning scheme (Feature Maps [Koh84]) with a nonlinear approximator (Backpropagation [RHW86]). The resulting learning system is more stable and e ective in changing environments than plain backpropagation and much more powerful than extended feature maps as proposed by [RS88, RMS89]. Extensions of our method give rise to active exploration strategies for autonomous agents facing unknown environments. The appropriateness of our general purpose method will be demonstrated with an example from mathematical function approximation.

[1] Ellis Horowitz,et al. Fundamentals of Computer Algorithms , 1978 .

[2] Geoffrey E. Hinton,et al. Learning internal representations by error propagation , 1986 .

[3] Teuvo Kohonen,et al. Self-Organization and Associative Memory , 1988 .

[4] Fujio Yamaguchi,et al. Curves and Surfaces in Computer Aided Geometric Design , 1988, Springer Berlin Heidelberg.

[5] Yves Chauvin,et al. A Back-Propagation Algorithm with Optimal Use of Hidden Units , 1988, NIPS.

[6] Helge Ritter,et al. Extending Kohonens Self-Organizing Mapping Algorithm to Learn Ballistic Movements , 1988 .

[7] David Haussler,et al. What Size Net Gives Valid Generalization? , 1989, Neural Computation.

[8] Klaus Schulten,et al. Topology-conserving maps for learning visuo-motor-coordination , 1989, Neural Networks.

[9] Christian Lebiere,et al. The Cascade-Correlation Learning Architecture , 1989, NIPS.

[10] Sebastian Thrun,et al. Adaptive Look-Ahead Planning , 1990, ÖGAI.

[11] Gerald Farin,et al. Curves and surfaces for computer aided geometric design , 1990 .

[12] Sebastian Thrun,et al. Planning with an Adaptive World Model , 1990, NIPS.

[13] Sebastian Thrun,et al. On Planning And Exploration In Non-Discrete Environments , 1991 .

[14] Michael I. Jordan,et al. Task Decomposition Through Competition in a Modular Connectionist Architecture: The What and Where Vision Tasks , 1990, Cogn. Sci..