What is machine learning and why is it so important?

Let’s start with a question.

What is the first image that comes to your mind after hearing artificial intelligence?

This article is one of my Article in Percept:

یادگیری ماشین چیست و چرا اینقدر مهم است؟

Most of people probably remembered Hollywood movies like Terminator or Oblivion or anything else fantasy like artificial intelligence is about to destroy the next generation of humans. But in fact, artificial intelligence is present for decades; Like your email spam!

Let us first identify the task of two terms that you hear a lot in the field of artificial intelligence and may have been mistaken for failure: 1) artificial intelligence, 2) machine learning,

Artificial Intelligence is a science similar to mathematics or biology that studies the methods of building intelligent programs and machines that have creative problem-solving.

Machine learning is a branch of artificial intelligence science that is able to create a problem in your operating system that can be solved without explicitly programmed programming using learning algorithms. In ML there are several algorithms (such as neural network algorithm) to solve the problem.

Why use machine learning?

Let’s consider the email spam program. What would you be surprised if you were to do this algorithm using conventional and direct programming techniques?

You will probably need to create a very long and complex list of rules and templates manually, each with its own algorithm for detecting spam emails. And this list must be constantly updated and all this must be done by the person. The idea of ​​such a world is a bit strange right now, isn’t it? If that were the case then maybe Google’s a much bigger team just for email spam.

On the other hand, spam filtering based on machine learning, you automatically learn that words and phrases are good for spam emails, and each user has extracted new patterns, even though personal emails can be ignored or so-called flagged. And reconstruct the method of detecting personal spam emails for that personal user.

As you can see, this program is much shorter and probably more accurate.

One of the other places where machine learning is most felt are issues that are very complicated with conventional methods, or there is no specific method. Imagine you want to write a program that recognizes the numbers one and two by sound. To do this, do a heavy programming with the size of the wavelength increase of these two words. Now imagine you want to recognize millions of words in different languages ​​this way to write a program that works in crowded environments. If this program was to be written this way, make sure that none of our phones have confirmed the ability of the voice assistant, why the implementation of this application with this method has been so effective. At present (at least so far) the best way to do this is with algorithms that learn from the data themselves.

Finally, machine learning will help people to learn better! You satisfied the machine learning algorithms and learned the results and patterns found. For example, applying learning methods in financial markets has given us a better understanding of trends, market patterns, and financial issues that have not been converted before, and it has done us a better job than that.

Using machine learning techniques to explore larger volumes of data needed to help discover patterns that were not previously detected. This is called data mining.

In summary, machine learning applies to the following:

• Problems that existing solutions require the implementation of many rules and patterns: In the case that with the help of learning you better code your machine and the performance of the program is better.
• Complex problems for which there is no decent solution with conventional programming methods, while learning algorithms require solving such problems.
• The volatile environments in which the existing data is constantly changing are adapted to the new data in how the operating system learns fast.
• Extract knowledge about complex issues and high data volumes

I hope you enjoy this post. The next article will talk about different types of machine learning methods.