Years ago by colleagues egg on turn type, from a focus on academic research related doc transformation of GIS programmer with engineering thinking now of course it is the development of deep learning algorithm engineer direction of natural language processing.
Go to work since February 3, realized with python practical cases a structured text extraction, scattered learned during various python programming, network infrastructure nerve, deep learning framework, etc., be simple into the door, especially CNN, different channel looked N times ha ha.
Because the epidemic, many network courses are open free courses, pay close attention to the time the car! To sum up, the current rush to get started mainly in these areas
1, Python Learning: After a white stage, now basically has started, gradually transition to higher-order
2, theoretical learning algorithms (mathematical foundation, the algorithm basis): This part of the basis used to have, but less practice, while practicing side can consolidate
3, deep learning framework for learning: tensorflow is currently on a net class, before the adoption of tf.keras simply into the door, with entry keras feeling is good, at least with a smile inside. Baidu have not tried the paddle, the future can try. After Take this as a framework for learning DL Sa
4, machine learning combat: There are a number of cases on the Microsoft Web site, you can practice
5, deep learning combat: with a lot of cases there are different frameworks of many NLP direction, as well as combination of book buying tensorflow
The road is long Come, happiness and earth ......
Finally recommend this entry written by zero-based depth study, very good, to understand too well:
Zero-based entry-depth study (1) - Perceptron
Learning Basics zero depth (2) - units and a linear gradient descent
Zero-based entry-depth study (3) - neural networks and back propagation algorithm
Zero-based entry-depth study (4) - convolution neural network
Zero-based entry-depth study (5) - Recurrent Neural Networks
Zero-based entry-depth study (6) - When the length of the memory network (LSTM)
Zero-based entry-depth study (7) - Recurrent Neural Network
