We are excited to showcase our latest achievement: the successful implementation of our first Printed Circuit Board (PCB) Analog Oscillatory Neural Network (ONN). In this implementation, each oscillator is based on a phase-locked loop (PLL). By coupling the analog PLL-based oscillators capacitively, we have effectively emulated Ising spin interactions and achieved negative coupling coefficients.
We would like to acknowledge the great work of our TU/e EE bachelor student, Vincent Cootjans, who has dedicated his project to designing, implementing, and testing the 3×3 PLL-based ONN both on breadboard and PCB. Not only has Vincent achieved successful implementation, but he has also benchmarked several network topologies on solving the max-cut problem.To provide you with a glimpse of this great project, we have included some pictures showcasing Vincent’s work.