ANN weight optimization using Cat Swarm Optimization

This toolbox updates the weights of ANN using CSO method.

ANN weights optimization using CSO has better ability to reach global minima than gradient descent method. This package was developed to predict Sea Surface Temperature anomaly (SSTA) time series for specific lead times. Results of prediction of SSTA with CSO and gradient descent was compared and found that CSO gives 20 to 40% improvement in root mean square error.