Open PhD Position

Hardware-Software Co-Design for Energy-Efficient On-Device Learning

Edge computing on mobile and wearable devices will is promising to democratize artificial intelligence to realize many useful applications in healthcare and robots to enhance the life quality of human beings. However, running deep neural networks (DNNs) in resource-constrained edge devices is difficult, especially in the learning of algorithms. This project will develop both efficient DNN learning methods and correspondingly optimized digital/mixed-signal hardware accelerators to realize the goal of energy-efficient on-device learning. The student will be fully funded with the TU Delft salary rate. The student is expected to have experience in training DNNs and digital/mixed-signal circuit design. Tape-out experience will be a plus.

Drop me an email to chang.gao at tudelft.nl if you are interested.