Fog forecasting using self growing neural network "CombNET-II"-a solution for imbalanced training sets problem
Anto Satriyo Nugroho, S. Kuroyanagi, Akira Iwata · 2000
Proposes a method to solve a problem that comes with imbalanced training sets which is often seen in practical applications. We modified the self growing neural network CombNET-II to deal with the imbalanced condition. This model is then applied to practical application which was launched in the '99 Fog Forecasting Contest sponsored by Neurocomputing Technical Group of IEICE, Japan. In this contest, a fog event should be predicted every 30 minutes based on the observation of meteorological conditions. CombNET-II achieved the highest accuracy among the participants and was chosen as the winner of the contest. The advantage of this model is that the independency of the branch networks contribute to an effective way of training and the time can be reduced.