My main research interest lies in the intersection of deep learning and neuroscience. On one hand, inspired by the brain, I develop new deep learning models to circumvent the difficulties that current deep learning models are facing. On the other hand, using deep learning techniques, I try to unravel the secrete of the brain about how it processes sensory information. One important ongoing project is neuron tracing in mouse brain.
My lab also conducts research on various applications of deep learning such as traffic sign detection, face detection, image segmentation, medical image analysis, music generation, etc. In recent years we have conducted many researches on the robustness of neural networks. An interesting application is the creation of "invisible cloaks" for person detectors based on neural networks.