We are seeking a highly skilled and motivated candidate to investigate and implement unconventional integrated platforms to implement spiking neural networks that rely on fast gain dynamics. Concepts of the in-memory computing will also be investigated and developed for this same platform and for the first time. This PhD research is strongly simulation, fabrication and characterization oriented and will lead to high impact results in the domain of optical self-learning.
Research challengesYou will be working within the newly granted Photon Delta National Growth Fund investment. See here for more information:
https://www.photondelta.com/. Within this extensive project framework, we will focus on mapping neural network models on free-space-optics setup for frontier neural network research and accelerators, as well as - as addressed in this position - miniaturize these concepts in photonic integrated chips. The project is also among the first one to suggest an unconventional platform to enable self-sustained spikes for lossless spiking neural networks. This research is strongly fabrication -oriented and will enable a whole new class of spiking neural networks, which could lead to highly efficient adaptive optical engines.
Photonics can play an important role in computation by designing and developing compact, energy-efficient, low latency and high-capacity computing engines. Several on-chip computing paradigms have already been proposed based on the use of light, but scalability is still a problem. A different approach needs to be explored and realized in this program. Within this project, the Photonic Neural Network Lab within the ECO group at TU/e aims to use straight-forward deposition techniques for interfacing gain-materials to photonic integrated waveguides and push their co-integration. Different materials and ligands will be explored in collaboration with the chemistry department to augment the system of brain-inspired functionalities, for the first self-sustainable spiking systems.
In this research program, the PhD student will explore the co-integration of gain layers on photonic integration platforms, for the implementation of miniaturized optical engines and of the first self-sustained spiking mechanisms. He/She will implement novel memory mechanisms, and push compactness of the structure that still allow self-sustainability. Will investigate scalability and benchmark with state-of-the-art AI chips. This research will enable a whole new class of self-reconfigurable photonic devices, which will lead to highly efficient parallel optical processing engines and high impact journals.
The teamThe PhD position is based within the Electro-Optical Communication Systems (ECO) group, which is part of the Eindhoven Hendrik Casimir Institute (EHCI) and of the Eindhoven AI System Institute (EAISI). The ECO group has about 73 members, 55 of which are PhD students and PDs, and 18 are staff members, focusing on photonic technology for communication and computation.
The EHCI has five dynamic and ambitious research groups, which are closely cooperating: a systems group, a photonic integration technology group and three materials research groups, with a predominant focus on neuromorphic and quantum computing is the ideal ecosystem where to develop this research. The EAISI brings together all AI activities of the TU/e. Top researchers from various departments and research groups work together to create new and exciting AI applications directly impacting the real world in collaboration with representatives from the industry.