In this project we will develop nanophotonic reservoir computing as a novel paradigm for massively parallel information processing. Reservoir computing is a recently proposed methodology from the field of machine learning and neural networks which has been used successfully in several pattern classification problems, like speech and image recognition. However, it has so far mainly been used in a software implementation which limits its speed and power efficiency. Photonics could provide an excellent platform for such a hardware implementation, because of the presence of unique non-linear dynamics in photonics components due to the interplay of photons and electrons, and because light also has a phase in addition to an amplitude, which provides for an important additional degree of freedom as opposed to a purely electronic hardware implementation. Our aim is to bring together a multidisciplinary team of specialists in photonics and machine learning to make this vision of massively parallel information processing using nanophotonics a reality. We will achieve these aims by constructing a set of prototypes of ever increasing complexity which will be able to tackle ever more complex tasks. There is clear potential for these techniques to perform information processing that is beyond the limit of today's conventional computing processing power: high-throughput massively parallel classification problems, like e.g. processing radar data for road safety, or real time analysis of the data streams generated by the Large Hadron Collider.