Development of the picture recommendation system using neural network.

Every day computer vision technologies are moving forward, new operation principles of neural networks are emerging, which allow to get more effective results and build deeper and more useful dependencies. The scope of recommender systems is evolving, becoming more universal and useful. In this paper, we will highlight the topic of image recommendations to users, both using the collaborative filtering method based on preferences of similar behavior, characteristics and interests of people, and using image feature extraction using various neural networks and further searching for “neighboring” objects by these features. In this paper, a description of the principles of operation of methods for the implementation of recommender systems is given, the architectures of the used neural networks were also described in detail. The results of this work can be useful in many areas related to the use of media content by users. Further development of work in this direction involves the improvement and optimization of written classes, the addition of new ideas and functionality. It is also possible to effectively supplement the recommendation functions with new algorithms in the field of convolutional neural networks, over which researchers conduct continuous work all over the world.