Am I understanding the difference between a neural network and a computer program correctly?

Question

Neural networks

Good evening! I am writing an article about a neural network and the following question arose: "What is the difference between a neural network and a simple computer program?". Do I understand correctly that the program will produce a result that is programmed, and you can drive another command into the neural network that was not programmed into it, it will think and give a certain answer?
PS I know that there is a lot of information on the Internet, but I just want to confirm my thought

Answer 1

The program has a thinker inside. He thinks faster than a man, otherwise why is he needed?
A thinker can work either exclusively on the logic and mathematics developed by programmers, or he can also use a neural network trained by programmers.
For some tasks, logic and mathematics are more reliable, better and simpler, and also easier to reconfigure.
For some other tasks, dry logic either does not do well, or requires a lot of programmers' work.
And for some of these some tasks, instead of a logical-mathematical block, it is possible to insert something like a kaleidoscope into the thinker, into which you can shine a flashlight and get a wonderful picture. Shine from a different angle - a different picture will turn out. In a kaleidoscope, there is no explicit logic or mathematics, there are mirrors, colored stones ... but everything is according to the laws of physics. So it is in the neural network - everything seems to be according to the laws of computer science, but at the same time - mirrors and glass.
Something like this...

Answer 2

No. A neural network is just a mathematical model that cannot do anything on its own.
A program that uses neural networks applies neural networks to input parameters and produces a result.
Machine learning is used to "program" a neural network.

Answer 3

An indiscreet question - with such an understanding of the topic, WHY write an article "about a neural network"? What information will be carried? To whom will it be useful?

Answer 4

Error in the question. It's like looking for the difference between a car and a vehicle.

Answer 5

If it's oversimplified.
The program is a clear sequence of actions If-Then-Otherwise (If high, then step back, otherwise jump).
The neural network is a complex fit to the answer, the answer is never exact. Even if he is 100% correct, then this is an accident. In a primitive version, the NN is the arithmetic mean between the examples on which this NS was taught.
If the network was taught to “walk” on commands, and you tell it to “take off”, then it will also give an answer, but it will most likely be absurd, but not obviously erroneous, i.e. will look like the truth.

Answer 6

A neural network is a mathematical model.
A program is a combination of instructions and data for hardware.
In any program, you can use different mathematical models, for example, using one random number generator, you can provide the user of the program with different distributions: normal, uniform, etc.
It is possible to write a program that at the input transforms the input data into a form that is convenient to use with a certain implementation of the neural network, and at the output - into a form convenient for the user of the program.
A neural network can be trained and can be used. The program that implements the neural network will remain the same, regardless of what happens to the neural network, whether it is learning or working properly.
In one program, you can implement many neural networks, a whole pipeline, although in most cases this is not practical, but theoretically possible.