A deep learning model is overfitting (good on training, poor on new data). Which action will improve generalization to new data?

Overfitting occurs when a model learns the training data too well, including noise, leading to poor performance on unseen data. Increasing the size of the training dataset provides the model with more diverse examples, helping it learn generalizable patterns rather than memorizing specific instances. Decreasing the number of input features (feature selection or dimensionality reduction) simplifies the model, making it less complex and reducing its capacity to overfit the training data. Conversely, reducing the training dataset or increasing features would likely exacerbate overfitting.