（1）基础知识
dlbook_cn 中文翻译版https://github.com/exacity/deeplearningbook-chinese
英文在线学习：http://www.deeplearningbook.org/
CNN_book http://lamda.nju.edu.cn/weixs/?AspxAutoDetectCookieSupport=1
CS224d-lecture lhttp://cs224d.stanford.edu/syllabus.html
（2）博客：
https://cloud.tencent.com/developer/article/1092091
（3）ocr 相关论文：
attention-based_google 2017-8-20
论文地址：https://arxiv.org/pdf/1704.03549.pdf
论文实现代码：https://github.com/tensorflow/models/tree/master/research/attention_ocr
论文解读博客：https://zhuanlan.zhihu.com/p/44572066
论文涉及的技术：cnn+AM+rnn autoregress 做loss函数而不是ctc
An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition 2015-07
论文地址：https://arxiv.org/pdf/1507.05717v1.pdf
论文解读：https://www.cnblogs.com/lillylin/p/6035815.html
论文实现：https://github.com/keras-team/keras/blob/master/examples/image_ocr.py
论文翻译：https://www.jianshu.com/p/9a3114e91c34
Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks 2013-12
论文地址：https://static.googleusercontent.com/media/research.google.com/zh-CN//pubs/archive/42241.pdf
论文解读：https://juejin.im/post/5c04e342f265da6165015391
依据论文延申的 验证码识别：https://github.com/nladuo/captcha-break/blob/master/weibo.com/weibo.com-process.ipynb
实验环境：https://colab.research.google.com/，可以打开ipynb文件，在线训练。
