Light-Neuro: En neuromorf protonisk plattform för accelererad beräkning

Purpose and goal:

The Light-Neuro project, a KTH and Conscience collaboration, aims to develop a neuromorphic protonic platform using lithium niobate thin film. KTH designs photonics circuits, while Conscience manufactures devices. This tech mimics neural networks for accelerated computing, achieving GHz switching and rapid info transmission. At TRL2, it´s set for further development, benefiting Sweden´s high-tech research and sustainability goals. This innovation could revolutionize computing and industries.

Expected results and effects:

This collaboration combines KTH´s nanophotonic expertise and ConScience´s nano-fabrication skills to create fast EO neuro-system using LNOI technology. The team aims to develop a light-speed classifier for image recognition, advancing neuromorphic computing and reducing power consumption. They´ll pioneer photonic chip platforms, explore non-volatile circuit elements, enhance scalability, and create novel architectures.

This partnership promises to revolutionize integrated photonics and computing, removing barriers to energy efficiency and processing speed. Approach and implementation:

Month 1-2: System Design (WP1) A team from ConScience and KTH will create a detailed system architecture for the LNOI neuromorphic circuit. Month 2-6: Fabrication (WP2) ConScience will manufacture, based on the WP1 design, in their facilities. Month 2-6: Testing (WP3) Both parties will test the circuit for performance and reliability.

Month 5-6: Validation (WP4) After testing and software development, the circuit will be validated through pattern recognition tests. Month 7-8: Report (WP5) & Project Management (WP6) ConScience and KTH will compile a technical report on the system