Data Sampling
Domain: Data & Datasets
🎯 Motivation
Selecting a subset of data for training or evaluation. Different sampling strategies for different needs.
📋 Concrete Example
Sampling: from 1M articles, randomly select 100K for training. Or stratified sample: ensure equal representation from each category. Strategy matters.
🔗 Analogy
Taking a spoonful of soup to taste - you sample a small part to represent the whole. Data sampling selects representative subsets for training.