Feature Scaling normalizes or standardizes input variables so they contribute proportionately to distance or gradient calculations. Common methods include min-max scaling and standardization (z-score). It can greatly improve the performance of many algorithms, including gradient-based and distance-based models.