In the context of artificial intelligence (AI), generalization refers to the ability of a machine learning model to provide accurate outputs for inputs it has not previously seen during its training phase. The goal of a well-trained model is not just to memorize the specific examples in the training data, but to learn underlying patterns or principles that allow it to apply what it has learned to new data.
Generalization is a critical concept in machine learning. If a model performs well on its training data but poorly on unseen data, it is said to be „overfitting“. Overfitting means that the model has learned the training data too well, including its noise and outliers, and fails to generalize to new data. On the other hand, if a model performs poorly on both the training data and unseen data, it is said to be „underfitting“. Underfitting implies that the model has not learned enough from the training data to make accurate predictions.
The balance between learning from the training data and being able to generalize to unseen data is a fundamental aspect of machine learning. Various techniques, such as cross-validation, regularization, early stopping, and model selection, are used to improve the generalization capability of a model.