Explain gradient descent and its variants.

Gradient descent optimizes model parameters by iteratively moving in the direction of steepest descent of the loss function.\n\n• Batch GD: Uses entire dataset, stable but slow\n• Stochastic GD: Uses single sample, fast but noisy\n• Mini-batch GD: Compromise between batch and stochastic\n• Advanced: Adam, RMSprop, AdaGrad with adaptive learning rates\n\nExample: For neural networks, use Adam optimizer with learning rate 0.001, batch size 32. Monitor loss curves, adjust learning rate schedule, and use gradient clipping to prevent exploding gradients.