Convolution in the end what role? How to do feature extraction?

[Study notes]

After the previous basis for learning neural networks, and finally into the core of our chapter, convolutional neural network (CNN, Convolutional Neural Networks). Many students learned a long time convolution neural network, but there has been a fundamental problem not get to know the most, he also knows how to do convolution, and also know how to do pooling, just I do not know why I did so? Mark -to-win @ Mark java community: has been foggy. General videos or books never mention this, always one step tall on a variety of moving map, showing how volume how volume. No way, this task falls on my shoulders, and I try to put it in layman's language to speak to understand it! Curly plot in the end what role? How to do feature extraction? Pooling product and how to do it, most importantly, why.

Reprinted from the original article: https://blog.csdn.net/qq_44639795/article/details/100603849

Guess you like

Origin www.cnblogs.com/haima1949/p/11512844.html