Differentiable neural computer

A differentiable neural computer (DNC) is a machine learning model consisting of a a neural network that combines the learning and pattern-recognition strengths of deep neural networks with the ability to retain information in complex data structures.