What is artificial intelligence

Artificial intelligence is divided into 2 types: weak and strong. Weak AI is also called narrowly focused, because it can only perform tasks within a certain framework. These are all developments based on AI technology that exist today. Strong artificial intelligence will be able to solve any problem in an unlimited range of areas.

AI technology can be implemented in different ways. One of the ways is neural networks. The neural network is built on the same principle as the neural networks in a living organism, hence the name. In the body, the network connects nerve cells - neurons, they form the nervous system. And in an artificial neural network, simple processors are used - computational elements that connect and interact in the same way.

Unlike conventional algorithms, neural networks are capable of learning from experience. Neural networks analyze and identify connections between input and output data, summarize data and form solutions to problems. To enable neural networks to function in this way, machine learning methods are used. Moreover, in the case of neural networks, such training requires a lot of computational resources.

What you can teach a neural network depends on the input data. The more data, the better the training will be. You can teach a neural network to distinguish some objects from others, to compare and predict. Learning a neural network is similar to teaching children, when they are shown a picture and said: "This is a cat." In the case of neural networks, they receive a lot of such pictures with explanatory labels and learn to recognize individual elements, which they can then combine.

The input image gets into some kind of filtering system. The filters in it are different in size and complexity of the elements that can be recognized - each has its own set of features. The image is filtered many times in this system. When many elements are recognized, the neural network makes a forecast: with such and such a probability this object is a person.

You can get advice on creating the artificial intelligence necessary for you (your business) by contacting the AI ​​developers https://data-science-ua.com/consulting/

NVIDIA engineer Robert Bond developed an algorithm that turned on garden sprinklers when neighbors' cats wandered in and ruined his garden. To determine if it was a cat, he used a system based on the Caffe neural network: it identified cats from video footage from cameras. When the camera recorded the change in the situation, it took 7 photos. The photographs were analyzed by a neural network: if there was a cat in the pictures, then the network included sprinklers.

Report this Page