JSIII - PHOTONICS FOR ARTIFICIAL INTELLIGENCE
Antonio Hurtado, University of Strathclyde, Glasgow,United Kingdom (Co-Chair)
Bruno Romeira, INL Institute, Braga, Portugal (Co-Chair)
“Nowadays, there are clear global challenges associated to the vast amounts of energy we consume in processing Big Data using Artificial Intelligence (AI) that are putting our industries and societies at risk. AI operate at high-power penalty since they rely on intensive data-driven deep learning neural networks running in conventional computers. There is a demand for new computing paradigms able to run AIs at extremely low energy per bit budgets. Neuromorphic systems, that can mimic the way the brain process information, are among the most promising technologies. Photonics enables the design of energy-efficient computing approaches such as neuromorphic photonic computing, allowing massively distributed power-efficient architectures for parallel processing and enabling new AI applications as required for the Industrial 5.0 Intelligent era revolution. This symposium covers the status, prospects, and challenges of light-based computing for AI taking advantage of new materials, devices, architectures, software, algorithms and simulation tools.
Device, circuit; architecture design; analysis and optimization for neuromorphic photonic computing systems; Emerging materials for devices of neuromorphic photonic computing importance; Hardware photonic accelerators for machine/deep learning; Photonic reservoir computing; Novel neuromorphic photonic computing systems; On-chip learning and inference, learning algorithms and optimizations; Complexity and scalability of neuromorphic photonic computing; Emerging technologies for brain-inspired nanophotonic computing and communication.