Verfahren, Datenverarbeitungsvorrichtung, Computerprogrammprodukt und Datenträgersignal

Details

Veröffentlichungs-Nr.

EP4040312

Aktenzeichen

EP21156128

Anmeldetag

9. Februar 2021

Veröffentlichung

10. August 2022

Rechtsraum

EP

IPC

G06F16/906

Abstract

The invention relates to a method for providing a balanced training dataset for training a Machine Learning model, a data processing device, a computer program product, a data carrier signal, and a method for selecting representatives of data records within a dataset.A method for providing a balanced training dataset (20) for training a Machine Learning model (30), comprising: receiving a multi-class dataset (10) containing data records (11) of at least one majority class (12) and at least one minority class (13), representing the data records (11) of the at least one majority class (12) by using at least one content-based representation of the data records (11), k-means clustering of the data records (11) of the at least one majority class (12) based on the at least one content-based representation, wherein the number k of clusters is set in consideration of a number and/or size of the at least one minority class (13), selecting the data record (11s) closest to the centroid of each cluster as representative of the respective cluster, aggregating the selected data records (11s) of the at least one majority class (12) and the data records (11) of the at least one minority class (13), providing the aggregated data records (11, 11 s) as training dataset (20).

Anmelder

Firma

Volkswagen Aktiengesellschaft

Land

🇩🇪 Deutschland

Fachgebiete