I will introduce the research effort to improve the efficiency of Deep Neural Networks at any different levels.

Model efficiency: the designed Neural Networks for various computer vision tasks and achieved > 10x faster speed and lower energy.

Data efficiency: to develop an advanced tool that enables 6.2x faster annotation on point cloud. also domain adaptation  to utilize simulated data, bypassing the need for real data.

Hardware-efficiency: co-designed neural networks and hardware accelerators and achieved 11.6x faster interface.

Design efficiency: the process of finding the optimal neural network is time-cosuming .  automated neural architecture search algorithms are discovered, using 421x lower computational cost, models with state-of-the-art accuracy and efficiency.