"Human brains are powerful at interpreting speech and visual signals. Can we get inspiration from human brains to build highly capable computational neural networks to achieve similar performance? The answer is yes. In this talk, I will introduce our research progress on developing brain-inspired deep learning algorithms, which have similar and interesting mechanism such as long-term memory engagement, adaptive information flow regulation, and selective attention, etc. Our simulation results on realistic image data show that such brain-inspired mechanism largely advances the state-of-the-art performance."