One of the reasons for the expansion of the use of artificial intelligence is deep machine learning, thanks to which computers in some areas have surpassed people’s abilities. Deep learning through multiple layers of non-linear transformations distinguishes the desired features [1 ]. The Deep Blue supercomputer from IBM defeated the world chess champion Garry Kasparov, despite the fact that until recently computers were unable to solve trivial problems from a human point of view, i.e. natural language processing or recognition of objects in photographs. Both of these activities are performed outside our consciousness. In 2006 Geoffrey Hinton and other researchers presented a deep learning algorithm that recognizes handwritten figures from the MNIST database [2 ]. It contains a training set of 60,000 examples and a test set of 10,000 examples. The solution is considered to be a breakthrough because, based on the test set, the precision of the solution was over 98%.

#dropout #pooling #convolutional-network #kernel #deep learning