ABSTRACT
The SAM architecture is a novel neural network architecture, based on the gross architecture of the cerebral neocortex, for combining unsupervised learning modules. When used as the high-level behavioral mechanism of an agent, the architecture enables the agent to learn the functional semantics of its high-level motor outputs and sensory inputs, and to acquire high-level and complex behavioral sequences by imitating other agents (learning by ‘watching’, or learning from a coach). This form of learning involves the agent attempting to recreate the sensory sequences it has been repeatedly exposed to. The architecture should scale well to multiple motor and sensory modalities, and to more complex behavioral requirements. Finally, insofar as it is based on the architecture of the cerebral neocortex, the SAM architecture may also help to explain several features of the operation of the cerebral neocortex.
