Domestic read books about the depth of learning, you can take a look at "the depth of learning principles and application of practice," describes the depth of relatively simple learning principles (Chapter 3 and 4 of 19). Only describes the "neural network" and "convolutional neural network", the other types of depth neural network (such as automatic encoder, recurrent neural network) not involved.
Caffe deep learning open source tools framework and source code analysis of the content in more detail (Chapter 5 of 60), with emphasis on the practical part convolution neural network. Describes the application scenario with eight chapters eight image recognition, such as handwritten digit recognition, face recognition, face recognition, recognition of age.
The last part of the summary of the "deep learning defect" is quite frank. For example, the authors note: "deep learning of the final stage turned into a boring, mechanical, parameter adjustment and time-consuming work," and since that: "engaged in depth study and research, such as the author researchers rarely in theory argumentation, deduction has made an important contribution. "the authors point sober and honest praise!
Depth study for beginners, engineers and graduate students, which detailed experimental guidance valuable. A clerical error found: page 45 in FIG. 4-7, "S3" should be "S2".
"Deep Learning: Principles and Application of" Chinese version of the PDF, with a table of contents and bookmarks.
Learning network disk download: http://106.13.73.98
