Machine Learning Video Library

Abstraction: Caltech indexed video segments covering core ML topics

Key points:

• Organized by topic: bias-variance tradeoff, gradient descent (batch and SGD), neural networks (backpropagation, multilayer perceptrons), SVMs (hard/soft margin, kernel methods)
• Covers VC dimension theory including growth functions, shattering, and distribution-free bounds
• Includes regularization techniques: weight decay, Tikhonov, augmented error, early stopping
• Validation section covers cross-validation (leave-one-out, 10-fold), model selection, and data contamination
• Topics include overfitting, RBF networks, logistic regression with cross-entropy, and ensemble learning
• Netflix Competition used as motivating example for SVD and stochastic gradient descent

Connections: Caltech · Machine Learning · Neural Networks · Support Vector Machines · Gradient Descent · Vc Dimension

Source: http://work.caltech.edu/library/index.html