In writing this article, I remembered the fifth class, at that time I was able to simplify and be well-versed in Sudakshya Angk. The class teacher was fascinated by me and said, "Your head is not a computer!" Although at that time without much of the word "computer", I did not know anything more than that, I just knew that all the wonderful things can be done. But even though I did not know it, now I know- people's brains are actually the most amazing things. The computer does not have intelligence like ours, this machine just commands the direction of the halo.
But if you want to make the computer intelligent and clever, how will it be possible? Scientists have run away for this question, and have discovered artificial intelligence, machine learning, deep learning technology. After many decades of pursuit, teaching computers was a very difficult thing to teach the computer. However, with the help of neural networks , millions of artificial brain cells like humans can do exactly the work of the people. Neural networks are also known as artificial neural networks - which are part of deep learning. So what is it actually and how it works? Let's know the details ...
Brain versus computer
All the creatures in your imagination, such as mammals, birds, reptiles, fish, amphibians, etc., all have brain or brain. But people's brains are completely different than everyone else. Although not too big to draw, it helps to talk, think, solve problems, and find the important and amazing parts of the human body. A typical brain contains small cells like 100 billion (which may contain between 50 billion and 500 billion) - called neurons .
Each neuron has a cell body and a lot of the denadraitasa (Dendrites) (bahanakare cells took up the input kosadeha) and an Ask (Axon) (data output from the barrier cells that) is. Neurons are extremely small, with about 100 neurons in each millimeter. Only 10 percent of the cells in the brain neurons throughout the US, and the rest of the cells into the glial cells (Glial Cells) or niuragliya (Neuroglia) say they niuranake support and protect the neurons, and the power transmission.
On the other hand, the computer has a very small switch instead of the brain cell, which is called a transistor . The current technology's latest micro-processor has more than 2 billion transistors installed. Normal micro-processors contain 50 million transistors, and these things are placed together on a circuit, which is 25 millimeter square.
So this is the basic structure of human brain and computer. The computer is a metallic box and it works on binary numbers and the brain is a living thing that works on feelings and memory-but this is not the only difference between computer and brine. The main difference is that the computer and the brain work in a completely different way. Transposors are easily installed on computer processors, each transistor is usually connected to the 2-3 transistors, it is logical gates.(Logic Gates). But neurons in the brain are extremely complex, and are connected to each other in parallel. Each neuron is connected to about 10,000 neurons. So computers and brains completely differentiate and work for this different structure of hundreds of millions of computer transmitters which are simply fitted in parallel and 10-100 times more cells than those in the brain, which are complexly interconnected. Computers are specially designed to preserve large amounts of data (which is meaningless to the computer itself), and computer programs are installed to tell the computer how to process these data.
On the other hand, the brain is late to learn things, to remember, to practice repeatedly, and learn different things in a variety of ways, can create new learning methods itself - in this way creativity is seen in human work. The human brain can remember any patterns, then they can match, and see the known things differently.
What would have been like, if the computer was working like a blind person? And here's the neural network.
What is the neural network?
If you want to make a computer like a brein, it will be necessary to build a brain in the brain according to the general idea. In other words, there should be innumerable number of brain cells that will be related to each other - so that the system can be used to identify the pattern, learn something, and take concrete decisions on a subject like humans. The neural network will be the most interesting thing, that the computer does not have to create a program to do any work, it will learn to work automatically, just like us.
But to tell the truth, it is not possible to design a computer in the brain, but it is not a hardware, but using software like normal transistors is trying to behave like a computer-billion brain cell computer. The computing we are currently doing and the program that we operate, its output is worth not only to us, but to computers. Computer never knows what you're doing with her. The task that you do not let the computer complete - the computer will calculate them, change the binary, process the numbers, sort them in different patterns and you will see the output. For example, data storage, you have music, videos, photos, and know how many preserves it. Every data you have is different, but your phone or any computing device will have your music, photos,
So basically the neural network is a computing system that can understand everything like a human, learn from it, and gradually improve any flaws. The computer will see a dog photo as a live photo of a dog just by not seeing it as just a data (1/0) - and this is the neural network. And since the neural network is operated by software, it is not like a brain bone, so we call this whole process artificial neural networks or ANNs. Although it works like a human brain, it is not a brain but it is not a brain.
How does the neural network work?
See the process of doing it or the problems that lie inside it are a bit confusing, in which there will be many terms that will need to be discussed separately. So try to understand the whole subject with simple language and adequate examples, rather than technical ones.
There may be thousands of artificial neurons in a standard neural network, their units are called Units. These units are arranged randomly with each other and each one is connected to each other, just like a network. Some of them are units that accept different types of information, such as we see an object, try to color, understand drawing etc. - these units are called input units . These units help identify, process, or process neural networks. On the other side of the network, the output unit - which reveals exactly what computer got from the bustling things. Not only this, but the input unit and output units are in the middle of the field of the hidden unit., Which is composed of most of the network's artificial neurons. So it was found that there are three types of units in it, their group is called a layer. Each layer unit is interconnected with each unit and provides the latest result or output only after processing data from each other. Each unit's connections with each other are drawn by a number-it is called weight . If a unit agrees with the information of another unit, then the weight is positive and if the consent is refused then the weight is negative.
Now let's talk about how this works, in any two ways in the neural network, data is flowing. When the computer learns something about it, one kind of data flows and after receiving the instruction, when the computer is taken into action, then one kind of data is flowing. When the computer is taking an education, it requires frequent feedback to fix his learning, that he is learning the right and not learning the wrong. For example, if you have written a spelling mistake in front of a teacher, then you can improve it, and we always need feedback to learn something. Think you're doing balloons with airguns. You saw the gun on the balloon and made it to the balloon, but the shot went on an over. Now, for the second time when you do the suit, keep in mind the positions and targeting of the previous times. Now surely the gun will do a lower suit, Kenona knows your brain, She could not put her up high in the previous times. In this way you will be able to make the right target by refusing to try or correct a mistake.
Neural networks also receive similar education. Every time it receives information from the input unit. Suppose the input to the computer shows a cat photo. It preserves these information, such as what it looks like, what size, where the eyes are, where the face is, etc. Now every cat is no different, so it is necessary to show many cat photos to teach the network. Now if the computer is told by showing another photo, it is not a cat or a dog. But first it will receive its data input unit and send it to the redirect unit. The hidden unit has lots of information from the previous cat photo, so it starts to analyze the input. Suppose a unit will look at the nose, type, color, etc. If it matches the cats, then it will provide positive weight, Another unit may observe the eye and another may look at the tail. So this output unit will provide output based on positive and negative results. If the output is wrong, then the computing system should tell that it has given a wrong result, the computing system will not make that mistake again. Just as we learn from mistakes.
Now think you have taught neural networks with some chairs and table pictures. But now the whole new model chairs presented the picture of the table before, which was never shown before. Now the computing system will take the idea from the previously presented chairs tables (as the man does) and split the new chair tables into separate categories. But what kind of education you have given is also important. It will work with your previous education and learn new things every time.
Now a man was told, "Tahmid, look at the chair tables in front of you". In this way, however, no computer can be told that computer input must be done in binary - because no matter how many people try to do it, it is not a man. We know that switching on or off for each input of the computer (transistor ON / OFF) The Neural Network will respond to you only negatively or positively. Suppose you input the table chair image, 1) does it have the space to rear? 2) What is the surface? 3) Is there a soft mattress? 4) Can you sit here comfortably for a long time? 5) Can you put many things on it? Now the computer will answer you "in the table", yes, no, yes, yes, no, or the binary answer will be 10110 and will not answer the table, yes, no, no, Yes or banner 01001 That is, while learning, it will see the table as 01001 and see the chair as 10110, and it can also understand the chair or table.
Last word
Improvement of the neural network, if possible, will change the entire computing technology. Besides, Neural network is currently being used for automobile autopilot, credit card transaction, radar scanning control, handwriting recognition, voice recognition, email spam. Maybe in the future, using this computing technology will make it possible to create a very intelligent robot, which will behave like a human. However, today's issue was really complex, sometimes some technical terms came up, but still I tried to do my best to express it in simple terms.
Hopefully, it feels great to learn something new-you too much. Tutorial down to today's topic or any questions you may have about any tech, and of course share the post.
Congratulations @ziutso! You have completed the following achievement on Steemit and have been rewarded with new badge(s) :
Award for the number of upvotes
Click on the badge to view your Board of Honor.
If you no longer want to receive notifications, reply to this comment with the word
STOP
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Congratulations @ziutso! You have completed the following achievement on the Steem blockchain and have been rewarded with new badge(s) :
Award for the number of upvotes
Click on the badge to view your Board of Honor.
If you no longer want to receive notifications, reply to this comment with the word
STOP
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Congratulations @ziutso! You received a personal award!
Click here to view your Board
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Congratulations @ziutso! You received a personal award!
You can view your badges on your Steem Board and compare to others on the Steem Ranking
Vote for @Steemitboard as a witness to get one more award and increased upvotes!
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit