Social collaborative filtering recommendation method based on federal learning

Abstract

The invention discloses a social collaborative filtering recommendation method based on federal learning. The social collaborative filtering recommendation method comprises the following steps that 1, a central server selects user clients participating in local training; step 2, E rounds of local training are carried out on the user client, and in each round of local training process, a model loss function is calculated through a matrix decomposition module, a user comparison module and an article comparison module; 3, after the E rounds of local training are completed, a gradient which finally needs to be uploaded is obtained through a gradient protection module, and the gradient is uploaded to a central server for gradient aggregation; step 4, repeating the step 1 to the step 3 to obtain a fully trained user embedding vector and an article embedding vector; and 5, taking the user embedded vector and the article embedded vector as input of a score prediction module to obtain an article sequence recommended to the user. According to the method, the influence of the data non-independent identical distribution problem in federated social recommendation on the model recommendation performance can be effectively relieved.

Description

Technical field [0001] The present invention relates to the field of artificial intelligence interaction technology, in particular a recommended method of social collaborative filtering based on federated learning. Background [0002] As a data-driven application, the recommendation system generally trains the recommendation model centrally by collecting the user's personal information and interaction records (browsing, scoring, etc.), capturing the user's interest preferences, and thus generating recommendations to the user. However, there are huge privacy concerns about the centralized storage of these user information. [0003] With the introduction of privacy protection bills such as GDPR, the issue of privacy protection in the recommendation system has become a key research issue. As a paradigm of decentralized machine learning, the advantages of federated learning in protecting privacy have gradually been valued by scholars at home and abroad. Federated learning first trai...

AI Technical Summary

Problems solved by technology

However, they did not pay attention to the impact of non-IID data on the federated recommendation model, which will lead to a severe performance degradation of the federated social recommendation system

Because the update direction of the user's local model may be different from the update direction of the global model, the result of model parameter aggregation may deviate from the global optimal result.

[0004] In summary, although the current federated social recommendation method can solve the problem of user privacy and security and the cold start problem of new users, it ignores the impact of data non-independent and identical distribution on model performance

Images