Decision model training method, prediction method and device based on longitudinal federation learning

Abstract

The invention relates to a decision model training method, prediction method and device based on longitudinal federated learning. The training method comprises the steps that a training sample set participating in training and a training feature set are randomly determined; at each node of the current decision tree to be trained, determining node information of the node, and generating a first split gain set; receiving a split sample set generated by at least one second terminal; generating a second split gain set based on the split sample set and the label data; determining a splitting feature of the node based on the first splitting gain set and the second splitting gain set; and generating the decision model based on the plurality of trained target decision trees. According to the method and the device, the problem that the label data is leaked in the transmission process can be avoided, the safety of the node data of each participant is improved, and the data privacy of each participant node is protected.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular to a decision model training method, prediction method and device based on longitudinal federated learning. Background technique [0002] Federated learning refers to the calculation process in which multiple data owners can perform model training and obtain the final model without the original data being localized, and ensure that the gap between the model effect and the aggregated training effect is small enough; according to the data distribution, Federated learning can be divided into horizontal federated learning, vertical federated learning, and federated migration learning. [0003] For vertical federated learning, since different features of the same sample belong to different training participants, and the sample labels and features are also in a separate state, that is, the training participants have the same sample space and different feature...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by this application is to provide a decision model training method, prediction method and device based on vertical federated learning, which does not need to introduce third-party nodes in the federated learning process, so that it will not be affected by the introduction of third-party nodes. Brings additional data security risks; and for participant nodes with tag data, the tag data will not be exported locally, thereby avoiding the problem of tag data being leaked during transmission and improving the security of each participant node data , which protects the data privacy of each participant node

Images