Gradient Boosting vs. Random Forests, Batch Normalization, and SGD Momentum

Explain the following machine learning concepts at the level of a quant or engineer who understands the basics but wants the real intuition: **(a)** What is the key difference between Gradient Boosting and Random Forests? When would you choose one over the other? **(b)** What is batch normalization in deep learning, and why does it help training? **(c)** What is momentum in stochastic gradient descent, and why does it speed up convergence?