Unsupervised Representation Learning for Contextual Bandits

We propose a model for optimal decision-making on arbitrary, complex modalities, with applications to healthcare, robotics, HCI, and general artificial intelligence.