Do you need it?

[sumaiyakhatun27]

But classical machine learning has significant drawbacks - the data for processing and examples of solutions are given to the machine by a person. In other words, in classical machine learning, the computer performs a large number of tasks, but does not form these tasks on its own. The concept of "deep learning" assumes that the machine itself creates functionality for itself, as far as this is possible at the current moment in time.


The very concept of “depth” as applied to machine learning implies modeling netherlands email list multi-level abstractions (“layers”) and translating them into data. The more of these “layers” (“depths”), the more highly intelligent tasks a neural network can perform. Moreover, it performs them without human assistance. In essence, “deep learning” is the first and quite big step towards artificial intelligence. Neural networks are systems that, albeit primitively, can think , i.


e. create new discoveries from the data that they have created for themselves. The mechanism of multi-level abstractions (“layers”) used in deep learning is very similar to the mechanism of learning of a human child. Just as a child first learns sounds, then individual words, and only then sentences, a “deep” neural network moves from simple (superficial) abstractions to more complex ones. The deep learning revolution Throughout 2012 and 2013, Google quietly but methodically bought up existing companies and startups in the field of deep learning.