Objectives BACKUP tackles the challenges of realizing a hybrid neuromorphic intelligent network where photonic circuitry provides both the core computing power as well as the interface to biological neurons and electronic circuitry provides both the required I/O control and the platform where algorithms (i.e., deep learning networks) emulate the biological network. After realizing a massive reservoir-computing photonic network based on a complex topology and a neuron/integrated photonic circuit interface where light controls both the topological connections (synapses) and the activity of the biological neurons, BACKUP will develop:

1. dynamic memories in photonic integrated circuits (PIC) using reservoir-computing network (RCN);

2. time-series forecasting for both noisy and chaotic inputs;

3. an artificial network (software-implemented) which emulates the physical hybrid network;

4. a hybrid network where biologic and artificial networks collaborate to jointly solve computation tasks;

5. artificial memory engrams to understand the cellular base of memory and the neuronal plasticity during learning;

6. hybrid circuits controlling neuronal hyperexcitability and seizure-induced plastic changes.

All these are extremely high-risk activities with extremely groundbreaking objectives. However, strong scientific and societal motivations – the limit of standard computers and the impact of neurological disorders on the population – motivate this research.

Recent progresses in photonics, computer science, deep learning, time series analysis, optogenetics and neurophysiology support my ambitions.