Focuses on human‑in‑the‑loop learning and data quality: modeling annotator bias, bias‑aware label aggregation, and trustworthy ML pipelines using probabilistic approaches, evaluated across synthetic and real‑world datasets.