APPARATUS AND Method for EVALUATING MACHINE LEARNING MODEL

Provided are a method and an apparatus for monitoring accuracy of a provisioned ML model. The method comprises the steps of: deciding to make an NWDAF including an AnLF check the accuracy of the provisioned ML model; collecting data to monitor the accuracy of the ML model; and reselecting or retrain...

Full description

Saved in:
Bibliographic Details
Main Authors LEE SOOHWAN, SHIN MYUNG KI, SUNG JIHOON
Format Patent
LanguageEnglish
Korean
Published 18.07.2023
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC COMMUNICATION TECHNIQUE
ELECTRICITY
PHYSICS
TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHICCOMMUNICATION
WIRELESS COMMUNICATIONS NETWORKS
Online AccessGet full text

Cover

More Information
Summary:Provided are a method and an apparatus for monitoring accuracy of a provisioned ML model. The method comprises the steps of: deciding to make an NWDAF including an AnLF check the accuracy of the provisioned ML model; collecting data to monitor the accuracy of the ML model; and reselecting or retraining the ML model which is determined to have decreased accuracy based on the collected data. AnLF를 포함하는 NWDAF에게 프로비저닝된 ML 모델의 정확도를 확인할 것을 결정하는 단계, ML 모델의 정확도를 모니터링하기 위한 데이터를 수집하는 단계, 그리고 수집된 데이터를 바탕으로 정확도가 저하된 것으로 판단된 ML 모델을 재선택 또는 재훈련하는 단계를 통해 프로비저닝된 ML 모델의 정확도를 모니터링하는 방법 및 장치가 제공된다.
Bibliography:Application Number: KR20230004173