What is artificial intelligence? In layman's terms, it is to make machines think like humans. There is no need to explain this much, because we are already familiar with artificial intelligence through various science fiction movies. What everyone is interested in now is-how to implement artificial intelligence?
Since the term artificial intelligence was first coined in the summer of 1956, scientists have tried various ways to achieve it. These methods include expert systems, decision trees, inductive logic, clustering, and so on, but these are all false intelligence. It wasn't until the advent of artificial neural network technology that the machine had true intelligence.
Why are the previous methods all false intelligence? Because we humans can clearly understand their internal analysis process, they are just a large and complex program; artificial neural networks are different. It is a black box inside, just like our human brain, we don't Knowing its internal analysis process, we don't know how it recognizes human faces, nor how it defeated the Go world champion.
We just built a shell for it, just like humans, we just gave birth to a child. We don't know what he thinks! This is the terrible aspect of artificial intelligence, because in the future it may feel that we humans should not live in this world and destroy us; for this reason, many security associations have been established in the world to prevent artificial intelligence.
Artificial neural networks are created by being inspired by the structure of the human brain, which is the fundamental reason for its true intelligence. In our brain, there are billions of cells called neurons, which are connected into a neural network.
Artificial neural network is exactly imitating the above network structure. The dendrites of human brain neuron cells receive multiple stimuli of varying intensity from the outside, process them in the neuron cells, and then convert them into an output.
The simpler the structure of the brain, the lower the IQ. Single-celled organisms have the lowest IQ. Artificial neural networks are the same. The more complex the network, the more powerful it is, so we need deep neural networks. The depth here refers to the more layers, the more complex the neural network constructed.
The process of training a deep neural network is called deep learning. After the network is built, we only need to be responsible for continuously inputting training data into the neural network, and it will constantly change and learn on its own.
Let's say we want to train a deep neural network to recognize cats. We just need to keep feeding the cat's pictures into the neural network. After the training was successful, we arbitrarily took a new picture, and it could determine whether there was a cat in it.
But we don't know what his analysis process is and how it determines whether there are cats in it. Just like when we teach children to know cats, we bring some white cats, tell him that they are cats, and bring him some black cats, tell him that this is also a cat, and he will constantly learn the characteristics of cats in his mind. In the end we brought some cats and asked him, he will tell you that this is also a cat. But how did he know? What is the analysis process in his mind? We have no way of knowing
Through the study of this article, we know that real artificial intelligence can be realized through artificial neural networks. In the next article, I will explain the neural network in detail for everyone.