Author: Zen and the Art of Computer Programming
1 Introduction
The rise of artificial intelligence has attracted great attention. With the advancement of technology, the algorithms of machine learning are becoming more and more diverse, and the application of deep learning (Deep Learning) is more extensive. Since deep learning has become the hottest technology today, I think it is very necessary for beginners who want to learn this technology to master its basic concepts and algorithms.
In this article, I will use easy-to-understand language to guide you to understand some basic knowledge about neural networks, convolutional neural networks and backpropagation algorithms. These are the contents we need to understand and master, which can help readers deepen their understanding of deep learning.
As a senior programmer and software architect, I value the fundamentals of computer science and am very good at teaching students. Therefore, I hope that through this article, I can help more people get started with this technology quickly and help them establish the correct knowledge system.
2. Explanation of basic concepts and terms
2.1 Neurons
A neuron in the human brain consists of an axon, dendrites, and multiple cortical cells. Axons are responsible for receiving external input, and dendrites transmit signals to other neurons; there is a complex connection process between the two, and neurons can secrete many different types of neurotransmitters, which are important for information processing have an important role. Each neuron has a threshold, and only input signals exceeding this threshold are passed on to the next layer, otherwise the neuron is activated and neurotransmitters are released, allowing the signal to pass through the neuron.
As shown in the figure, axons and dendrites are the two most basic components, but they can also have multiple other parts, for example, synapses with cone-shaped structures, filamentous synapses, etc., and connections between different levels , The processing method is also different.
We know that the basic working principle of a neural network is: to obtain input data from the input layer, after processing by multiple hidden layers, and finally output