Neuromorphic Circuits for Novel Devices

This project is part of the H2020 MSCA-ITN MANIC (grant ID: 861153).

Successes in deep learning show that the paradigm of neuromorphic computing is very attractive. However, current technology is based on the Turing/von Neumann architecture, requiring extensive communication and an excessive amount of energy for computing. The human brain performs pattern recognition tasks with a fraction of the power needed by supercomputers for similar tasks. In this multidisciplinary project, new (memristic) materials and architectures for neuromorphic computing will be investigated in order to develop materials that can learn.

This project focuses on the design of neuromorphic circuits for embedding the novel devices developed by the synthesis partners of this ETN. Key neural computational primitives will be targeted; their circuital implementation and integration with novel devices will be characterized and used as building blocks for the construction of neural networks.