Machine Learning Series Chapter 1

An introductory tutorial that uses Andrej Karpathy’s Micrograd to demystify core ML concepts—supervised/unsupervised learning, regression vs classification, loss functions, gradients, backpropagation, and gradient descent—then builds up neurons, layers, and an MLP to show how parameters and hyperparameters influence training. Through Python snippets and DAG visualizations, it demonstrates manual vs automatic backpropagation and a simple training loop that reduces loss, positioning Micrograd as a minimal, transparent tool for learning how neural networks learn.