Method and device for sampling unbalanced data under federated learning

Abstract

The invention discloses an unbalanced data sampling method and device under federated learning, and the device comprises a data monitoring module, a data balancing module, a local training module, anda central server. An unbalanced data balancing scheme based on mixed sampling comprises the steps: according to the unbalance proportions of different data sets, obtaining a balance data set based ona mixed sampling method combining a few synthesized samples and a clustering down-sampling integration method; in combination with data set real-time monitoring, automatically setting an unbalanced data set processed and timely updated in a federated learning scene; therefore, through an unbalanced data set processing method combining a data level and an integration mode, the capacity of the dataset is fully utilized, and automatic equalization and updating of the unbalanced data set are realized by detecting the data set through data change detection.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a sampling method and device for unbalanced data under federated learning. Background technique [0002] At present, the problem of unbalanced data under federated learning mainly relies on the participants to consciously carry out local data before joint modeling to ensure the quality of local data. There are three main methods to deal with the problem of sample imbalance, (1) to obtain a balanced data set by modifying the data set; (2) to reduce the tendency to the majority class by modifying the machine algorithm; (3) to modify the previous One of the two methods is combined with ensemble learning to obtain a powerful ensemble classifier: [0003] (1) Data-level approach. This method obtains a sample-balanced data set by undersampling, that is, deleting the majority category to make it the same number as the minority category, or oversampling, that is, adding the min...

AI Technical Summary

Problems solved by technology

[0008]Because in the real scenario of federated learning collaborative modeling, the data distribution of each participant is different, and the unbalanced data processing method adopted will also There are differences, and it is difficult to ensure the data balance of each participant

And when new data is added, it is difficult to ensure that all participants can update in time

Images