csMTL: a Context Sensitive Lifelong Learning System

csMTL, or context-sensitive Multiple Task Learning, is presented as a method of inductive transfer that uses a single output neural network and additional contextual inputs for learning multiple tasks. The csMTL approach is demonstrated to produce hypotheses that are equivalent to or better than standard MTL hypotheses when learning a primary task in the presence of related and unrelated tasks. The paper also describes a machine lifelong learning system based on csMTL for sequentially learning multiple tasks. The approach satisfies a number of important requirements for knowledge retention and inductive transfer; taking advantage of representational transfer for rapid short-term learning and functional transfer for long-term consolidation.