What are neural networks?
Posted: Wed Jan 22, 2025 6:53 am
About 50% of Russians are worried that neural networks will take their jobs. They say that if you don’t master a new profession, you will be out of work in the next couple of years. But this is far from the truth. Vladimir Kazakov, Product Manager for the Training Business Unit at MTS Link, told us what neural networks are capable of and what they are not.
Subscribe to RB.RU in Telegram
In the classical view, a neural network is a cluster india telegram database of neurons. The more of them and the more connections between them, the more complex the network and the more complex tasks it can solve. Such systems can consistently process and output information.
Because a neural network is not an algorithm with precisely defined “if…then…” formulas, it can work with data it has never encountered before. For example, no algorithm can explain the phrase “Glokaya kuzdra shteko budlanula bokra i kurdyachit bokrenko.” But a neural network can understand that “some specifically characterized female entity did something in a specific way to another male entity, and then began (and continues to do to this day) to do something else to its offspring (or a smaller member of the same species).”
By the way, neural networks are often called artificial intelligence. But this is not entirely correct: human intelligence is capable of learning new things, while a neural network was once trained and uses the existing data as a copy until it receives new ones. To put it simply, we can say this: AI is the same artificial intelligence as a car is an artificial horse.
There are many types of neural networks, but for convenience we can conditionally divide them into two groups: specialized and general.
Subscribe to RB.RU in Telegram
In the classical view, a neural network is a cluster india telegram database of neurons. The more of them and the more connections between them, the more complex the network and the more complex tasks it can solve. Such systems can consistently process and output information.
Because a neural network is not an algorithm with precisely defined “if…then…” formulas, it can work with data it has never encountered before. For example, no algorithm can explain the phrase “Glokaya kuzdra shteko budlanula bokra i kurdyachit bokrenko.” But a neural network can understand that “some specifically characterized female entity did something in a specific way to another male entity, and then began (and continues to do to this day) to do something else to its offspring (or a smaller member of the same species).”
By the way, neural networks are often called artificial intelligence. But this is not entirely correct: human intelligence is capable of learning new things, while a neural network was once trained and uses the existing data as a copy until it receives new ones. To put it simply, we can say this: AI is the same artificial intelligence as a car is an artificial horse.
There are many types of neural networks, but for convenience we can conditionally divide them into two groups: specialized and general.