When I first started learning deep learning, I was often intimidated by some proper nouns, and I didn’t know what they meant. Many professional terms are very simple in layman’s terms, so here I record the meaning of each word in my own words.
1. Convolution
Convolution is a basic operation in graphics, and the convolution operation is performed through a convolution kernel. It is easy to describe with a diagram.
2. Down sampling (under-sampling)
3. Pooling
In layman's terms, pooling is to make the picture smaller, in order to reduce the amount of calculation of the computer.
There are three commonly used pooling methods: max pooling, mean pooling and global max pooling.
4.batch_size
5. Gradient Descent
Find the lowest point of the loss function
6. Learning rate
The distance of each move
The learning rate should not be too small, nor too large