Method and device for constructing decision tree

Abstract

The embodiment of the invention discloses a method and device for constructing a decision tree. The specific embodiment of the method comprises the steps of sending a request for obtaining statisticalinformation of attribute information of a target category to at least one client; receiving the statistical information of the attribute information of the target category of the sample respectivelystored in each client; generating split point information according to the statistical information of the attribute information of the target category of the sample stored by each client; and sendingthe split point information to at least one client. According to the implementation mode, the content of information interaction is greatly reduced, privacy protection is achieved, and meanwhile the efficiency of the model is improved.

Description

technical field [0001] The embodiments of the present disclosure relate to the field of computer technology, and in particular to a method and device for constructing a decision tree. Background technique [0002] In the era of artificial intelligence, data is like the energy for the development of the industry. The core of artificial intelligence is to enable the algorithm to learn the corresponding model based on the given data. Without enough effective data, artificial intelligence cannot be efficiently applied to people's daily life middle. But on the other hand, the excessive collection of data has brought about a crisis of privacy and security. Therefore, on the premise of ensuring the security of personal privacy, how to use data reasonably and legally to provide people with efficient services has become a research hotspot. [0003] Decision tree is the most commonly used algorithm for machine learning. Unlike the uninterpretable nature of neural networks, decision ...

AI Technical Summary

Problems solved by technology

Although the existing joint tree building model is modeled on the premise of protecting user privacy, due to the distribution of data in different platforms, the heterogeneity of samples on each platform, the imbalance of data, and the differences in network capabilities have all caused tree building problems. The problem of high communication costs in the process directly affects the performance of the joint modeling tree model. In practical applications, the pressure on network communication is extremely high, which cannot well meet the needs of existing scenarios.

Images