Task decomposition through competition in a modular connectionist architecture
Robert A. Jacobs, Andrew G. Barto · 1990
A novel modular connectionist architecture is presented in which the networks composing the architecture compete to learn the training patterns. As a result of the competition, different networks learn different training patterns and, thus, learn to compute different functions. The architecture performs task decomposition in the sense that it learns to partition a task into two or more functionally independent tasks and allocates distinct networks to learn each task. In addition, the architecture tends to allocate to each task the network whose topology is most appropriate to that task, and tends to allocate the same network to similar tasks and distinct networks to dissimilar tasks. Furthermore, it can be easily modified so as to learn to perform a family of tasks by using one network to learn a shared strategy that is used in all contexts along with other networks that learn modifications to this strategy that are applied in a context sensitive manner. These properties are demonstrated by training the architecture to perform object recognition and spatial localization from simulated retinal images, and to control a simulated robot arm to move a variety of payloads, each of a different mass, along a specified trajectory. Finally, it is noted that function decomposition is an underconstrained problem and, thus, different modular architectures may decompose a function in different ways. A desirable decomposition can be achieved if the architecture is suitably restricted in the types of functions that it can compute. Finding appropriate restrictions is possible through the application of domain knowledge. A strength of the modular architecture is that its structure is well-suited for incorporating domain knowledge.