ELI-chan for Intermediate Features
We propose ELI-chan, an emergent local indicator mechanism, to model the representation and self-organization of intermediate features in the visual pathway. This model is motivated by the orientation specificity in the primary visual cortex. Our simulations of ELI-chan demonstrate that local indicators of the locations of intermediate features emerge, and they become the seeds for unsupervised learning and pattern recognition; ELI-chan predicts those portions of the input imagery where intermediate features potentially exist. Hence, ELI-chan can be used to define a set of intermediate features for adaptation, and the onwards processing in a hierarchical pattern recognition system.
Physical Sciences and Mathematics
Strategy and Organisation
International Joint Conference on Neural Networks: IJCNN '93 Proceedings, 25-29 October 1993, Nagoya
City or Country
TING, Christopher Hian Ann.
ELI-chan for Intermediate Features. (1993). International Joint Conference on Neural Networks: IJCNN '93 Proceedings, 25-29 October 1993, Nagoya. 2508-2511. Research Collection Lee Kong Chian School Of Business.
Available at: http://ink.library.smu.edu.sg/lkcsb_research/779