Data Balance
Domain: Data & Datasets
🎯 Motivation
Having equal or proportional representation across categories. Imbalanced data causes biased models.
📋 Concrete Example
Balance: 50% cat images, 50% dog images. Balanced. vs 90% cat, 10% dog. Imbalanced - model will be biased toward cats. Balance ensures fair training.
🔗 Analogy
A balanced diet - equal portions of protein, vegetables, and carbs. Data balance ensures the model learns all categories equally.