How to train your first neural network

learn to teach the neural network is much simpler than you think

The Main trend of the last few years certainly can be called neural networks, machine learning and everything connected with them. And there are serious reasons, because in recent years neural network surprise with their skills. Moreover, the neural network may have only one of their votes and to "revive" portraits of Dostoevsky and Marilyn Monroe, so she is still able to show, 20, 30 and even 50 years! Of course, all this does not a neural network — in the world there are many similar projects, which involved experts in Data Science.

How did neural networks

It All started with the attempts of scientists to bring the principle of operation of the computer to the mindset of the person. It took decades of research, and eventually this became possible with the help of neural networks — computer systems composed of hundreds, thousands or millions of artificial brain cells that can learn and act on the principle, very similar to the work of the human brain.

Of Course, impossible to say that a neural network is an exact artificial copy of the brain. It is important to note that the neural network is first of all a computer simulation: these networks are created by programming conventional computers, in which the traditional way are ordinary transistors, combined in a logical connection.

As a neural net generates a new photo

what are neural networks

Conventional artificial neural network consists of tens, hundreds, thousands, or even millions of artificial neurons. They are called blocks — they are arranged in layers, where each unit is connected to the next. There are blocks of input with which the neural network receives the information, and output units — they are responsible for the treatment result.

When the net is trained, samples of the information "fed" to her through the blocks of the input, and then get to the units output. For example, you can show the neural network a huge number of photos of chairs and tables, the maximum available explaining to her the difference between these pieces of furniture. And then ask her to identify the object in the picture, which shows a wardrobe. Depending on how effectively you have trained the neural network, it tries to relate what he saw to the category, based on its experience.

How to train neural networks

Neural networks are trained with "the method of error back-propagation". With it is possible to compare the output with the data expected to obtain and use the differences between these data for the change in connection between the blocks involved in the network. The more trained neural network, the faster it turns to reduce to zero the difference between the desired and actual results.

One of the machine learning models.

Once the neural network was trained using a sufficient number of examples, it reaches the stage where you can provide her a brand new set of input data, which she had never seen, and watch for her reaction.

The use of neural networks is not limited. So, they can search by picture or to act as a voice assistant — the same Alice already as close in its behaviour to the real person. Or calculate the probability of diseases, to locate tumors in the images, to fight fraud and so on.

is it Possible to learn to work with neural networks

Previously, this possibility was granted only to scientists, because experience in the field of neural networks and machine learning was too raw. But now any technology company generates huge amount of data to be processed, then based on it to optimize the business and to analyze the prospects. For this and other problems associated with neural networks and machine learning, we need experts in Data Science.

How do I become one? Independently make it almost impossible. This is a serious specialization, which requires interaction with those who are already working in this area. Therefore, the school data SkillFactory opens a new set at full . In this course, industry professionals, including employees of Yandex, and NVIDIA, are taught the intricacies of working, which are not written in textbooks.

All the teachers — experts in the field of Data Science

You can use this to learn the science of working with data from scratch, even if you never in life were engaged in programming. It allows you to obtain all the skills required expert in Data Science — from programming in Python, including in-depth study of Pandas for data analysis, to m, deep learning and data science. approximately 20% theory and 80% practice, since the only real examples opportunity to become a Pro in this field.

the course runs for 12 months

In the learning process, you will be able to create their own projects in the field of image recognition, NLP and scoring. Together with teachers and mentors will understand the details of the job and get the necessary feedback. In addition, SkillFactory help with employment and recommend to the internship in large companies. For example, graduates have the opportunity to work in Alfa-Bank, Bayer, Henkel, Sberbank and other leading organizations.

On completion of training a certificate is issued

Now and get 15% discount in training on promo-code Hi-news (valid until 15.02.2020). Kit soon will be over, so time to think not so much.