The architecture and self-learning method of hybrid neuro-fuzzy systems for big fuzzy clustering in on-line mode are proposed in this paper. The architecture of proposed system represents the hybrid of the fuzzy general regression neural network and clustering self-organizing network. During a learning procedure in on-line mode, the proposed system tunes both its parameters and its architecture. For tuning of membership functions parameters of neuro-fuzzy system the method based on competitive learning is proposed.
The paper analyzes the current state of development and application of recommendation systems, models and methods of construction of recommendation systems. It is shown that the most widely used method came into collaborative filtering. The method of fuzzy clustering is developed, which improves the accuracy of predicting ratings of products.
The problem of clustering vector data sets with missing values in some components is considered. The adaptive approach to clustering of data in situation then classes overlap is proposed. The basis of the approach is the using of the modified Kohonen maps with the neighborhood function of special kind.