Backpropogation Lab Report

ABSTRACT:

This paper presents a BackPropogation neural network for predicting earthquake in Japan with proper data regarding the earthquakes. These parameters are taken from Japan seismic catalogue included all minor, significant occasions and aftershock sequences in Japan. This information contains event information, time of the event, latitude, longitude, magnitude, root mean square and depth of the earthquake. These information are grouped together to form seismicity indicators , It is then used as inputs for BP neural network for learning and predicting the magnitude of earthquake. These scientifically registered indicators considered are grouped based on each events above 3.5 range , add up to number of occasions from 2010 to 2016, b-values, energy released from the occasion,

The longitude is the quantity of degrees east (E) or west (W) of the prime meridian which goes through Greenwich, England.

[3] MODELING BACKPROPAGATION ALGORITHM

The backpropagation algorithm is utilized as a part of layered feed-forward Artificial Neural Networks. Backpropogation is a multi-layer feed forward, supervised learning system in light of gradient descent learning rule.We give the algorithm cases of the inputs and outputs we need the system to register, and afterward the error (distinction amongst real and expected outcomes) is ascertained. The possibility of the backpropagation algorithm is to decrease this error, until the point when the Artificial Neural Network takes in the training data. The neural network model is shown in figure

The activation function is generated using the sum of the inputs xi multiplied by their respective weights wji.

The most well-known output function is the sigmoidal