🤖 Machine Learning

Scheme REPLs run in the browser. Python equivalents in collapsible blocks.

	Chapter
1.	What is ML?	Function approximation from data — supervised, unsupervised, reinforcement	🤖
2.	Linear Regression	Find weights that minimize squared error — normal equation or gradient descent	🤖
3.	Optimization	Follow the negative gradient downhill — learning rate, convexity, convergence	🤖
4.	Logistic Regression	Squash linear output through sigmoid for probabilities — cross-entropy loss	🤖
5.	Kernel Methods	Replace dot products with kernel functions to learn nonlinear boundaries	🤖
6.	Neural Networks	Stack linear layers with nonlinearities — universal approximation by composition	🤖
7.	Backpropagation	The chain rule applied layer by layer — how gradients flow backward through a network	🤖
8.	Regularization	Penalize complexity to prevent overfitting — L1, L2, dropout, early stopping	🤖
9.	Convolutional Networks	Exploit spatial structure — shared filters, pooling, translation equivariance	🤖
10.	Recurrent Networks	Process sequences by feeding hidden state forward — vanishing gradients and LSTMs	🤖
11.	Dimensionality Reduction	PCA finds the axes of maximum variance — compress data without losing structure	🤖
12.	Clustering	Group data by similarity — k-means, EM, and the bias-variance tradeoff in unsupervised learning	🤖

Neighbors