Gail A. Carpenter and Sai Chaitanya Gaddam report–November’s most viewed on Digital Common
“Biased ART: A Neural Architecture that Shifts Attention Toward Previously Disregarded Features Following an Incorrect Prediction,” by Gail A. Carpenter and Sai Chaitanya Gaddam was the most viewed article on BU Digital Common last month. The technical report was added to DIgital Common on November 11 and was viewed 42 times during the rest of the month.
Abstract: Memories in Adaptive Resonance Theory (ART) networks are based on matched patterns that focus attention on those portions of bottom-up inputs that match active top-down expectations. While this learning strategy has proved successful for both brain models and applications, computational examples show that attention to early critical features may later distort memory representations during online fast learning. For supervised learning, biased ARTMAP (bARTMAP) solves the problem of over-emphasis on early critical features by directing attention away from previously attended features after the system makes a predictive error. Small-scale, hand-computed analog and binary examples illustrate key model dynamics. Twodimensional simulation examples demonstrate the evolution of bARTMAP memories as they are learned online. Benchmark simulations show that featural biasing also improves performance on large-scale examples. One example, which predicts movie genres and is based, in part, on the Netflix Prize database, was developed for this project. Both first principles and consistent performance improvements on all simulation studies suggest that featural biasing should be incorporated by default in all ARTMAP systems. Benchmark datasets and bARTMAP code are available from the CNS Technology Lab Website: http://techlab.bu.edu/bART/.
Digital Common, the University’s open access repository hosts a wide range of materials including published articles from faculty, technical reports, theses and dissertations, learning objects, and items digitized by the Libraries. For additional information about the repository, contact Vika Zafrin (email@example.com), the Institutional Repository Librarian for the BU Libraries.