Discussion of article "Neural networks made easy (Part 12): Dropout" - page 2

[Dmitriy Gizlyk]

Dmitry Nazarov:

Dmitry, can you explain a little bit more about the window parameter in the creation of Neuronkey. Am I right to understand that this is the number of neurons sequentially (with a step) that are taken for one-step calculation? To save computational power?

In a convolutional network, window is the maximum size of the pattern. Step is the size of the step with which the window is shifted to search for a pattern. For example, we are setting up a network to search for a fractal. We set window = 3, step = 1. We check the first 3 candlesticks and get some estimated value of pattern checking. We move by 1 candle and check the next 3 candles and get an estimated value for them. Thus, going through the whole sample, we get a vector of correlation values with the desired pattern.

[Trader Fx]

Understanding that artificial neural networks are a computational model that has evolved from various scientific contributions that are recorded in the history of artificial neural networks.