ARTD: Autonomous Recursive Task Decomposition for Many-Class Learning

摘要: Many-class learning is the problem of training a classifier to discriminate among large number target classes. Together with dealing high-dimensional patterns (i.e. input space), many-class output space) major obstacle be faced when scaling-up systems and algorithms from small pilot applications full scale applications. The Autonomous Recursive Task Decomposition (ARTD) algorithm here proposed as solution learning. Example ARTD neural are also presented. In these examples, improvements in time shown range 4-fold more than 30-fold pattern classification tasks both static dynamic character.