Chloe I want to understand how neural networks learn from data. How are weights and biases adjusted during training to improve predictions? Can someone also explain concepts like backpropagation and gradient descent in simple terms?