Synthetic Data Generation Strategies for Pipeline Robustness
A comprehensive survey and analysis of synthetic data generation techniques, classifying them by data modality, generation method, and application purpose and exploring how synthetic data contributes to the resilience of ML pipelines against failure modes such as concept drift, noise, and adversarial attacks.