TY - JOUR

T1 - Statistical Process Monitoring of Artificial Neural Networks

AU - Malinovskaya, Anna

AU - Mozharovskyi, Pavlo

AU - Otto, Philipp

N1 - Funding Information: We acknowledge the support of the cluster system team at the Leibniz University Hannover, Germany, in the production of this work. The project is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - 412992257.

PY - 2024

Y1 - 2024

N2 - The rapid advancement of models based on artificial intelligence demands innovative monitoring techniques which can operate in real time with low computational costs. In machine learning, especially if we consider artificial neural networks (ANNs), the models are often trained in a supervised manner. Consequently, the learned relationship between the input and the output must remain valid during the model’s deployment. If this stationarity assumption holds, we can conclude that the ANN provides accurate predictions. Otherwise, the retraining or rebuilding of the model is required. We propose considering the latent feature representation of the data (called “embedding”) generated by the ANN to determine the time when the data stream starts being nonstationary. In particular, we monitor embeddings by applying multivariate control charts based on the data depth calculation and normalized ranks. The performance of the introduced method is compared with benchmark approaches for various ANN architectures and different underlying data formats.

AB - The rapid advancement of models based on artificial intelligence demands innovative monitoring techniques which can operate in real time with low computational costs. In machine learning, especially if we consider artificial neural networks (ANNs), the models are often trained in a supervised manner. Consequently, the learned relationship between the input and the output must remain valid during the model’s deployment. If this stationarity assumption holds, we can conclude that the ANN provides accurate predictions. Otherwise, the retraining or rebuilding of the model is required. We propose considering the latent feature representation of the data (called “embedding”) generated by the ANN to determine the time when the data stream starts being nonstationary. In particular, we monitor embeddings by applying multivariate control charts based on the data depth calculation and normalized ranks. The performance of the introduced method is compared with benchmark approaches for various ANN architectures and different underlying data formats.

KW - Artificial neural networks

KW - Change point detection

KW - Data depth

KW - Latent feature representation

KW - Multivariate statistical process monitoring

KW - Online process monitoring

UR - http://www.scopus.com/inward/record.url?scp=85171781969&partnerID=8YFLogxK

U2 - 10.48550/arXiv.2209.07436

DO - 10.48550/arXiv.2209.07436

M3 - Article

AN - SCOPUS:85171781969

VL - 66

SP - 104

EP - 117

JO - TECHNOMETRICS

JF - TECHNOMETRICS

SN - 0040-1706

IS - 1

ER -