Using the most advanced technologies in programming neural networks and computer training based on past periods
Neural Network Model for Flu Sickness Rate prediction.
We propose new quantitative methods utilizing deep learning for flu sickness rate forecast.
The accuracy of predictions according to test results is 85%.
The neural network consists of a group of receptors, 4 hidden layers and one output neuron.
The group of receptors is supplied with a segment of historical flu sickness data.
Each hidden layer of a neural network consists of n-neurons; each following layer is completely connected to the next one.
The function of activating hidden blocks ensures non-linearity of the network.
The network was trained by using the Back Propagation (BP) algorithm.
The output layer consists of one output neuron which is producing an appropriate sickness rate.
The node of the output level has a sigmoidal activation function.