Five different sources are reviewed to understand Concept Drift in neural networks.

1) https://www.nature.com/articles/s41467-024-46142-w - Empirical data drift detection experiments on real-world medical imaging data

2) https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2024.1330258/full - One or two things we know about concept drift—a survey on monitoring in evolving environments. Part B: locating and explaining concept drift

3) https://research.google/blog/learning-the-importance-of-training-data-under-concept-drift/ - Learning the importance of training data under concept drift

Then two research papers:

4) https://arxiv.org/pdf/2004.05785 - Learning under Concept Drift: A Review

5) https://arxiv.org/pdf/2203.11070 - From Concept Drift to Model Degradation: An Overview on

Performance-Aware Drift Detectors

These sources collectively explore the critical issue of concept drift in machine learning, which refers to systematic changes in data distributions over time that can degrade model performance. The "Nature Communications" excerpt details empirical experiments on real-world medical imaging data (chest X-rays) to evaluate data-based drift detection methods, finding that monitoring performance alone is often insufficient to detect such shifts. Complementing this, "Frontiers" provides a broader survey on monitoring, localizing, and explaining concept drift, particularly in unsupervised settings, and discusses how drift intensity and data dimensionality impact detection. The final "arXiv" papers offer comprehensive reviews of concept drift research, outlining a framework of detection, understanding, and adaptation, and classifying performance-based detection methods while also categorizing various types of concept drift (e.g., sudden, gradual, incremental, recurring) and their probabilistic sources.