From drift to adaptation to the failed ml model: Transfer Learning in Industrial MLOps

TL;DR

This paper evaluates transfer learning strategies for updating failed ML models in industrial MLOps, focusing on accuracy and computational efficiency through a case study on power plant data.

Contribution

It compares ensemble, all-layers, and last-layer transfer learning methods for model adaptation in industrial settings, providing practical insights for MLOps practitioners.

Findings

ETL outperforms other methods for 5-day batch updates.

ALTL is effective for large batch size updates.

Computational requirements vary with batch size and method.

Abstract

Model adaptation to production environment is critical for reliable Machine Learning Operations (MLOps), less attention is paid to developing systematic framework for updating the ML models when they fail under data drift. This paper compares the transfer learning enabled model update strategies including ensemble transfer learning (ETL), all-layers transfer learning (ALTL), and last-layer transfer learning (LLTL) for updating the failed feedforward artificial neural network (ANN) model. The flue gas differential pressure across the air preheater unit installed in a 660 MW thermal power plant is analyzed as a case study since it mimics the batch processes due to load cycling in the power plant. Updating the failed ANN model by three transfer learning techniques reveals that ETL provides relatively higher predictive accuracy for the batch size of 5 days than those of LLTL and ALTL. However, ALTL is found to be suitable for effective update of the model trained on large batch size (8 days). A mixed trend is observed for computational requirement (hyperparameter tuning and model training) of model update techniques for different batch sizes. These fundamental and empiric insights obtained from the batch process-based industrial case study can assist the MLOps practitioners in adapting the failed models to data drifts for the accurate monitoring of industrial processes.