ConScience receives Swedish government grant for photonic neuromorphic computing
Artist's impression of microfluidic neuromorphic device. Copyright ConScience AB
ConScience, a company specialising in microfabrication and nanofabrication, has announced that, together with partners, it has received two project grants totalling 10 million Swedish Krona (around £75000) from Swedish innovation agency Vinnova. One of the projects focuses on developing neuromorphic computing – specifically making artificial neurons for micro and nanoelectronics. These neurons mimic the way the brain works and opens the possibility for alternative computing architectures and AI.
The project is a collaboration between KTH Royal Institute of Technology and ConScience, and aims to pioneer development of photonic neuromorphic computing systems on a lithium niobate on Insulator (LNOI) platform, thus tackling the limitations of standard silicon-based technologies. Inspired by the "More than Moore" notion, this research seeks to transcend the restrictions of traditional von Neumann approaches through integrated photonic, stochastic, and neuromorphic computing methodologies.
According to ConScience, this approach promises superior energy consumption and bandwidth performance, particularly advantageous for machine learning applications due to their highly parallel architectures and efficacy in matrix multiplication operations that are responsible for the major part of the energy consumption in today’s AI systems. The company says that SOI platforms, despite their demonstrated utility in transceiver development, are energy-inefficient when used for optical analogue computing.
The energy required for tuning the weights in photonic integrated circuits (PICs) is a major bottleneck hindering the miniaturisation of optical systems. LNOI is a compelling alternative, offering low-loss propagation, low-energy electro-static switching, and the possibility of dense integration with detectors and sources, through hybrid integration onto a single chip platform.
Advanced nanofluidic imaging technology
The other project for which ConScience has received a grant is a collaboration with Envue Technologies, a spinout from Chalmers University of Technology, to develop nanofluidic scattering microscopy (NSM). ConScience is responsible for the nanofluidic chip design and fabrication, aiming to design and build a single-molecule nanofluidic device to enable the NSM detection method.
Envue is developing technology utilising nanofluidic chips as tiny optical sensors in combination with advanced light illumination and data analysis to detect and measure individual biomolecules in solution. Envue’s patented innovation estimates a biomolecule’s mass and size – two key parameters in life science – without the need for modifying processes of labels or advanced surface chemistry. The technology has the potential to disrupt the bioanalysis market by enabling rapid measurement of biomolecules with a greater reliability and efficacy than existing alternatives.
Joachim Fritzsche, CEO of ConScience, said: “We are very excited to receive this continued support from Vinnova which enable us to strengthen our portfolio of beyond state-of-the-art microfluidics and computing technologies.”