Optimization

The process of adjusting a model's parameters to minimize the loss function during training. In practice, this means finding the combination of weights that makes the model's predictions as accurate as possible. Gradient Descent is the most widely used optimization approach in deep learning.