Periodic Reporting for period 1 - SPIKEPro (SPIKING PHOTONIC-ELECTRONIC IC FOR QUICK AND EFFICIENT PROCESSING)
Reporting period: 2024-03-01 to 2025-02-28
Rapid advances in artificial intelligence technologies have led to powerful models and algorithms that have revolutionized many applications across all fields of science and technology. Deep learning performed within artificial neural networks has yielded new ways to process data, leading to sophisticated systems with impressive functionality and benefits. However, conventional computing hardware is reaching its limits in terms of energy efficiency and speed. A new approach to computing hardware is needed. Novel brain-inspired or neuromorphic chips working with biologically-inspired spiking neural networks have gained attention as they promise highly efficient ways to process data. Important research effort has been dedicated to develop such neuromorphic systems in electronic or photonic hardware separately, each with its drawbacks and limitations. SPIKEPro proposes a science-towards-technology breakthrough by combining low-energy electrical and photonic neurons into a joint spiking neural network on an integrated circuit.
SPIKEPro’s chip integration approach is based on a common technology platform, connecting ultrafast laser optical neurons with efficient electrical spiking diodes through non-volatile synaptic weights. This enables to simultaneously capitalise on the advantages of both electronics and photonics to deliver efficient and high-speed SNNs going beyond existing implementations. In addition to reducing the energy consumption per spike in the network, SPIKEPro will also develop novel learning strategies and algorithms able to work with reduced number of synaptic connections. These will be possible by exploiting the hardware parameters of the electrical and photonic spiking devices. The outcome of SPIKEPro will have lasting economic, societal and scientific impact. The project will bring ultra-fast and efficient neuromorphic hardware into the disparate fields of edge computing, sensor data processing, high-speed control and computational neuroscience.
SPIKEPro’s chip integration approach is based on a common technology platform, connecting ultrafast laser optical neurons with efficient electrical spiking diodes through non-volatile synaptic weights. This enables to simultaneously capitalise on the advantages of both electronics and photonics to deliver efficient and high-speed SNNs going beyond existing implementations. In addition to reducing the energy consumption per spike in the network, SPIKEPro will also develop novel learning strategies and algorithms able to work with reduced number of synaptic connections. These will be possible by exploiting the hardware parameters of the electrical and photonic spiking devices. The outcome of SPIKEPro will have lasting economic, societal and scientific impact. The project will bring ultra-fast and efficient neuromorphic hardware into the disparate fields of edge computing, sensor data processing, high-speed control and computational neuroscience.
In WP2, which deals with the development of the spiking components, UCL has optimized the fabrication process for the active laser material. With the help of a new annealing technique, we were able to achieve good optical characteristics in the fabricated wafers. UCL and TUE have defined the layer stacks for the electronic and photonic spiking devices and wafers for the electrical devices were successfully grown. TUE has performed simulations and arrived at an initial design of the photonic spiking device whereas the electronic spiking device design is finalized. Memristor process tests have been devised by HPE and TUE.
Within WP3, we work on the electronic-photonic interconnectivity. The consortium has developed modeling approaches for coupling of electronic and photonic nodes. USTRATH has performed experimental investigation into optical-electrical transitions between spiking nodes and demonstrated basic processing tasks with such. TUI has implemented the electrical spiking model into dynamic simulations of networks for reservoir computing, and has demonstrated good performance for time-series tasks. Pre- and postprocessing methods have been analyzed by TUI and USTRATH, that increase the memory capacity in such networks.
In WP4, USTRATH has investigated experimentally the operation of discrete neurons under multiwavelength optical inputs and in multi-modal regimes of operation. HPE is developing a system-level simulator that is compatible with common ML software frameworks and can connect to SPIKEPro's hardware models.
Within WP5, work has started to prepare measurements for technologies and systems of WPs 2-4. Activities include preliminary development of PCB assembly of chips. Additionally, USTRATH has started early experimental work with discrete devices to verify the processing potential of spiking devices and conducted numerical simulations to study the computational capabilities of networked neurons.
Within WP3, we work on the electronic-photonic interconnectivity. The consortium has developed modeling approaches for coupling of electronic and photonic nodes. USTRATH has performed experimental investigation into optical-electrical transitions between spiking nodes and demonstrated basic processing tasks with such. TUI has implemented the electrical spiking model into dynamic simulations of networks for reservoir computing, and has demonstrated good performance for time-series tasks. Pre- and postprocessing methods have been analyzed by TUI and USTRATH, that increase the memory capacity in such networks.
In WP4, USTRATH has investigated experimentally the operation of discrete neurons under multiwavelength optical inputs and in multi-modal regimes of operation. HPE is developing a system-level simulator that is compatible with common ML software frameworks and can connect to SPIKEPro's hardware models.
Within WP5, work has started to prepare measurements for technologies and systems of WPs 2-4. Activities include preliminary development of PCB assembly of chips. Additionally, USTRATH has started early experimental work with discrete devices to verify the processing potential of spiking devices and conducted numerical simulations to study the computational capabilities of networked neurons.
In the first period the consortium has established the foundation for developing a unified technology platform, integrating ultrafast laser-based optical neurons with energy-efficient electrical spiking nodes. We established designs for the electrical and photonic spiking neurons compatible with a joint platform, which marks a significant step beyond the current state of the art. We also gained novel insight into how such spiking neurons can be interconnected and best used to solve certain types of problems. The platform is identified as a key exploitable result. The consortium has identified two more key exploitable results, a simulation codebase describing hardware spiking neural networks and hardware-software co-design methods that relate to SPIKEPro's technology. Both will make it easy for other researchers in the future to utilize SPIKEPro's technology, leading to scientific and commercial impact.