What Is Data Augmentation?

Data augmentation is a process of enhancing or adding to existing data sets in order to increase their predictive power. This increase in predictive power can come from generating new data points that are statistically similar to the existing data set, or through the addition of features that increase the data set‘s complexity and richness. Data augmentation can also help reduce the effects of overfitting, which occurs when models are trained excessively on the same data set and memorize it, leading to poor generalization performance when tested with new data. Data augmentation plays an important role in deep learning, where large amount of data is required for effective training.