Integration of Results from Recognition Algorithms Applied to the Uranium Deposits

抄録

Data interpretation of electric logging can be performed using self-learning systems such as artificial neural networks (ANNs). Preliminary research shows that by using ANN we can achieve 52–73% of coincidence of interpretable data and experimental results. Therefore, it is necessary to analyze the possibility of using other classification algorithms, and that of using several classification algorithms simultaneously through a unified system (referred to as an integrator. These algorithms may improve the quality of recognition of individual species. The problem of developing a recognition system that combines several classification algorithms, also known as the integrator, is formulated here. A simple algorithm is developed for the learning and recognition of an integrator for the post processing stage; this enhances the recognition accuracy by 1–3%.