Yu Jianguo, Ph.D., studied in the man-machine interface laboratory of Huijin University. His master's degree is to use deep learning to integrate mouth data into the training of speech recognition model, without mouth data, thus improving the recognition rate. I also continued my doctoral program because I like research. I hope to share with you my thoughts and self-study experience in recent years. You can search YJango on Zhihu to find the shared content, or you can check out his life-long serial gitbook "Super Agent" about how the machine learns and how the human brain learns.

Yu Jianguo, the author of Super Agent, shared the content of understanding why "deep" network is better than "shallow" network and what tasks deep learning is suitable for, and tried to find out feedforward neural network, circular neural network, volume and neural network, residual network, pre-training, multi-task learning, end-to-end learning, automatic encoder, transfer learning, division, dropout, regularization and batch norm. Cut into deep learning in another way.