AI ※
※
Informed Model Update
AI ※
Informed Model Update
Category: AI
Update a model only when deemed necessary, for example, based on concept drift detection. When a model is in production it is common that is performance degrades over time. This can happen due to multiple reasons: the context has changed, there is new data, etc. Hence, it is common practice to regularly update the model. However, it is important to make sure that we only update the model when there is a real need to do so. This way, we avoid wasting unnecessary resources to train new model. One way to detect model degradation is for example, using concept drift detection.
Sources
- Poenaru-Olaru, Lorena, et al. "Retrain AI Systems Responsibly! Use Sustainable Concept Drift Adaptation Techniques." 2023 IEEE/ACM 7th International Workshop on Green And Sustainable Software (GREENS). IEEE, 2023.
- Poenaru-Olaru, Lorena, et al. "Sustainable Machine Learning Retraining: Optimizing Energy Efficiency Without Compromising Accuracy." 2025 11th International Conference on ICT for Sustainability (ICT4S). IEEE, 2025, pp. 100-111. ↗