Fast learning in networks of locally-tuned processing units Export

@article{citeulike:3408169, abstract = {We propose a network architecture which uses a single internal layer of locally-tuned processing units to learn both classification tasks and real-valued function approximations (Moody and Darken 1988). We consider training such networks in a completely supervised manner, but abandon this approach in favor of a more computationally efficient hybrid learning method which combines self-organized and supervised learning. Our networks learn faster than backpropagation for two reasons: the local representations ensure that only a few units respond to any given input, thus reducing computational overhead, and the hybrid learning rules are linear rather than nonlinear, thus leading to faster convergence. Unlike many existing methods for data analysis, our network architecture and learning rules are truly adaptive and are thus appropriate for real-time use.}, address = {Cambridge, MA, USA}, author = {Moody, John and Darken, Christian J.}, citeulike-article-id = {3408169}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1351136}, citeulike-linkout-1 = {http://dx.doi.org/10.1162/neco.1989.1.2.281}, doi = {10.1162/neco.1989.1.2.281}, issn = {0899-7667}, journal = {Neural Comput.}, number = {2}, pages = {281--294}, posted-at = {2008-10-15 11:06:12}, priority = {2}, publisher = {MIT Press}, title = {Fast learning in networks of locally-tuned processing units}, url = {http://dx.doi.org/10.1162/neco.1989.1.2.281}, volume = {1}, year = {1989} }

TY - JOUR ID - citeulike:3408169 L3 - citeulike-article-id:3408169 N2 - We propose a network architecture which uses a single internal layer of locally-tuned processing units to learn both classification tasks and real-valued function approximations (Moody and Darken 1988). We consider training such networks in a completely supervised manner, but abandon this approach in favor of a more computationally efficient hybrid learning method which combines self-organized and supervised learning. Our networks learn faster than backpropagation for two reasons: the local representations ensure that only a few units respond to any given input, thus reducing computational overhead, and the hybrid learning rules are linear rather than nonlinear, thus leading to faster convergence. Unlike many existing methods for data analysis, our network architecture and learning rules are truly adaptive and are thus appropriate for real-time use. IS - 2 JF - Neural Comput. SN - 0899-7667 AD - Cambridge, MA, USA EP - 294 TI - Fast learning in networks of locally-tuned processing units PB - MIT Press VL - 1 SP - 281 AU - Moody, John AU - Darken, Christian PY - 1989/// UR - http://dx.doi.org/10.1162/neco.1989.1.2.281 ER -