Federal learning training platform and method based on terminal equipment

Abstract

The invention discloses a federal learning training platform and method based on terminal equipment, and belongs to the field of artificial intelligence, and the platform comprises at least one terminal equipment which is used for receiving an initial model, and each terminal equipment comprises a data collection module, a data processing module, and a training module; a cloud platform is connected with at least one terminal equipment, the cloud platform comprises an issuing module, a receiving module and a model aggregation module, and the model aggregation module is used for aggregating model parameters of the terminal equipment to obtain a shared model. The terminal equipment further comprises a model reasoning acceleration module which is connected with the model aggregation module and is used for performing reasoning acceleration on the shared model based on the AI end side reasoning framework. The platform has the beneficial effects that based on a federated learning strategy, each terminal equipment can cooperatively perform model training without giving own data, the problem of data privacy leakage is avoided, meanwhile, model reasoning is accelerated based on an AI end-side reasoning framework, the recognition speed and precision are improved, and the recognition range is wider.

Description

technical field [0001] The present invention relates to the field of artificial intelligence, in particular to a federated learning training platform and method based on terminal equipment. Background technique [0002] Artificial intelligence is a very broad science, which consists of different fields, such as machine learning, computer vision, etc. In general, the data required for artificial intelligence will involve multiple fields. However, there are difficulties between data sources. In most industries, data exists in the form of isolated islands. Due to issues such as industry competition, privacy security, and complex administrative procedures, even data integration between different departments of the same company still faces challenges. There are many obstacles. In reality, it is almost impossible to integrate the data scattered in various places and institutions. Even if it is possible, the integration process will require huge costs. [0003] With the further de...

AI Technical Summary

Problems solved by technology

[0004] In the existing technology, traditional machine learning generally processes data and trains models centrally on a single machine or cluster, and then distributes the model from the cloud. It needs to rely on massive data to improve the accuracy of the model obtained from training and analysis. Although Large-scale mobile phone data has greatly improved the performance of machine learning algorithms, but it also brings greater challenges to the protection of personal privacy data, especially in mobile terminals, where personal privacy data security issues have become more prominent. Therefore, in response to the above problems, there is an urgent need to Design a federated learning training platform and method based on terminal equipment to meet the needs of actual use

Images