Matthew What is a learning curve in machine learning, and how does it help evaluate a model's performance during training? How can training and validation curves help identify problems such as overfitting, underfitting, and insufficient training data?