Table 1 MLOps life cycle. Detail of the different blocks and specific objectives of the proposed prototype
From: A novel botnet attack detection for IoT networks based on communication graphs
Blocks | Objectives | Project scope |
|---|---|---|
Value proposal | Define problem and importance | AIDS with lower computational cost |
Data sources | Identify main sources | Network data flow |
Prediction task | Model type to use | Stacked autoencoder |
Features | How to represent input | CGN |
Offline evaluation | Define methods and metrics | Accuracy and F1 score and MSE |
Decisions | How to use predictions | Generate alerts |
Making predictions | When and how | Batch periodically |
Collecting data | Cost of new data | No labeling required |
Build models | Frequency and cost | Periodic re-training |
Evaluation and monitoring | How to supervise | Metrics with human oversight |