Neural Networks - adjustable model in which way - this input function. It consists of several layers:
- input level, which contains the input data;
- a hidden layer containing the processing units, which are called neurons
- output level, consisting of one or more neurons, which output data are output across the network.
All nodes of adjacent layers are interconnected. These connections are called synapses. Each synapse has a designated rate at which data is transmitted across the synapse, multiplied. This ratio is called a weight (w [i] [j] [k]). The neural network of direct distribution (Feed-Forward Neural Network, FFNN) data distributed from inputs to outputs. Here FFNN example with one input layer, one output and two hidden layers.
Source : http://Ценовой прогноз с использован...; MetaTrader 4
Put BPNN.dll into MQL4 > Libraries.
BPNN_Predictor.mq4 and BPNN_Predictor_with_Smoothing.mq4 into MQL4 > indicators.