Recommendation system based on meta-learning and reinforcement learning

Abstract

Through methods in the fields of meta-learning, reinforcement learning and data mining, a recommendation system based on meta-learning and reinforcement learning is realized, an internal updating module and a meta-updating module model are defined and constructed, and two modules form a system model; the system model training process is as follows: a system model is trained, a recommendation strategy is generated for characteristic data input of primary user; an error is solved according to the strategy; by an optimization method mentioned in an internal updating process, optimizing parameters of a model through an optimization method mentioned in the internal updating process to obtain a next strategy, a final strategy is acquired according to a set internal updating step number, finallythe error is generated by inputting feedback of a user to recommended contents, then an initial model is derived and a new model is acquired by updating. After the model training is completed, the system receives the feature data ofa user, recommends push content to the user, and collects feedback of the user to the content.

Description

technical field [0001] The invention relates to the fields of meta-learning, reinforcement learning and data mining, in particular to a recommendation system based on meta-learning and reinforcement learning. Background technique [0002] At present, the recommendation system is almost everywhere. Many apps use the recommendation system, such as travel, shopping, video, news, social networking, etc. We can all see the shadow of the recommendation system, which is closely related to our daily life. The reason is that for users, facing massive amounts of data, they always hope to quickly find information that is interesting or valuable to them; for information producers, they always hope that their content can attract more customers, but different customers have different preferences, so for different users, there should be different recommended content. Although the recommendation system has many benefits for both enterprises and users, if the performance of the recommendati...

AI Technical Summary

Problems solved by technology

On the other hand, with the rapid development of meta-learning in recent years, what it needs to solve is how to achieve good results when learning in the face of few samples. Its combination with reinforcement learning has also achieved good results on some issues. effect, so we use reinforcement learning and meta-learning methods to solve the problem of recommendation learning to quickly learn user preferences

Images