Training Data Index

Constructing high-quality training datasets.

Notes

• Data Labeling — annotation workflows, weak supervision, and quality control for constructing ground-truth datasets
• Class Imbalance and Augmentation — oversampling, undersampling, SMOTE, and augmentation strategies for skewed label distributions
• Dataset Versioning — DVC-based lineage, data contracts, and reproducible data snapshots

Links

• ↑ ML Engineering
• Prev: ← 02 — Data Engineering
• Next: 04 — Feature Engineering →