1. Introduction to deep learning
Currently, deep learning is active in various parts of the world. Deep learning is at work in the smartphones that nearly everyone owns, in cars that drive themselves, and in the servers that power Web services. At this moment, deep learning is quietly performing its functions in places that many people do not notice. In the future, deep learning is bound to become more active. Introduce the path of my introductory deep learning and the bibliography of reading, for reference.
2. Books on the basic theory of deep learning (must learn):
"Introduction to Deep Learning: Theory and Implementation Based on Python": A must-have book for getting started, all of which are basic knowledge related to deep learning . There is no need to type the code in the book by yourself. It is recommended to read and master quickly!- "Mathematics of Deep Learning": I haven't read much about this book. After reading the above one, you can take a look at this one. This book mainly explains some mathematical knowledge in deep learning in a concrete way.
- Note: It is recommended to read the above two books within a week. If you have a certain foundation, you can read it in a few days. It is mainly to check for leaks and fill in vacancies.
3. Notes (check for omissions and fill in vacancies)
- "deeplearning_ai": You don't need to read all of them, just look at the explanation of the corresponding network first for what type of network model you want to study! Regardless of the number of pages in this book, there are not many knowledge points that I have extracted after reading it.
- "Basic Tutorial for Machine Learning": This book seems to have been taken off the shelf. There may be some small mistakes in it. At that time, I read this book mainly because I wanted to advance to the mathematical level when I was learning GAN. In fact, it seemed useless, so I didn’t need to read it if I didn’t have enough time. If you read the paper in the future and it involves some machine learning things, such as what the posterior probability is, you can look back.
4. Framework learning class
- eat_pytorch_in_20_days：https://github.com/lyhue1991/eat_pytorch_in_20_days
- eat_tensorflow2_in_30_days：https://github.com/lyhue1991/eat_tensorflow2_in_30_days
- Explanation: The above two are the dry goods of the learning framework, and they should not be out of date now. There are many processes used by the framework, and it is also recommended to learn by query.
5. Experience
- "Yousan AI-Deep Learning Open Source Framework Practice Guide V1.0_2020.6.20"
- "Three AI-Deep Learning Visual Algorithm Engineer Growth Guidebook-2020.5.29"
Six, finally
When you have a solid foundation, you can download the open source code and actually start debugging. Build less models and modify more! ! ! step by step. Pay attention to your own research field, it is best to go to Google Scholar every few weeks to check what is new, to ensure that your research is not behind. Pay attention to the oral papers of the four top conferences, you can read related papers, and learn more from efficient modules.