This tutorial uses Lasagne (a tool that can quickly build neural networks based on Theano):1, to achieve the construction of several neural networks2. Discuss the method of data augmentation3, discuss the importance of learning "potential"4. pre-trainingThe above methods will help improve our results.
This tutorial is based on your knowledge of neural networks, because the working principle of neural networks will not be discussed here.But here is some good information:1, an online book of deeplearning2.
1. There are two ways to classify libraries. If you are familiar with DLL under WIN platform, I believe it is not difficult to understand:
libraries can be used in three forms: static, shared and dynamic. Static library's code is connected to the application developed by the developer at compile time, while shared libraries are loaded only when the program starts to run. At compile time, library functions that need to be used are simply specified.
First of all, I'm a computer student. I entered for the Central South University. I didn't enter for the postgraduate entrance examination tutorial class. Listening to the postgraduate entrance examination class is a waste of time. Many students gave the most precious summer vacation time to the postgraduate entrance examination class. I must review according to my own situation. The following are only my own learning methods and procedures for reference only.
QuickLZ is a compression library that claims the fastest compression speed in the world and is also an open source compression library that complies with GPL 1, 2 or 3 protocols.
there is a QuickLZ test on QuickLZ's website:
Library Level Compressed size Compression Mbyte/s Decompression Mbyte/s QuickLZ C 1.5.0 1 47.9% 308 358 QuickLZ C 1.
Although it takes a lot of mathematical knowledge to develop artificial intelligence, it is much simpler if we only apply it simply.I share the source code of face recognition here. I have tested it myself.PS：Please create your own image folder.And install opencv extension in advance: pip install opencv-python
Copy content to clipboard
import cv2# # # Call Webcam to Detect Face and Take Screenshotdef CatchPICFromVideo(window_name, camera_idx, catch_pic_num, path_name):cv2.
upper_bound (): returns the first pointer in the checked sequence that is greater than the lookup value; lower_bound (): returns the first pointer in the checked sequence that is greater than or equal to the lookup value; 1, lower_bound；; usage: int t=lower_bound(a+l,a+r,m)-a Explanation: Find the element with the value of M in the [l,r) interval by halves in
topic meaning ccc cats, ddd dogs, vvv individual votes are given. Everyone may only like a cat, hate a dog, or like a dog, hate a cat.Ask how many people can be satisfied at most by removing some cats and dogs.
thinking It is not difficult to see that liking a cat, hating a dog and liking a dog, hating a cat may be mutually exclusive.We will also introduce a concept of "
K K-nearest Neighbor k-Nearest Neighbor algorithm
In the picture on the right, which class should the green circle be assigned, is it a red triangle or a blue square?If K=3, because the proportion of red triangles is 2/3, the green circle will be assigned to the red triangle class; if K=5, because the proportion of blue squares is 3/5, the green circle will be assigned to the blue square class.