Advanced Learning Algorithms

Course 2 notes on neural networks, decision trees, and practical advice for building machine learning systems in the Andrew Ng specialization.

Topics

Course 2 of 3 in the Machine Learning Specialization by Andrew Ng (DeepLearning.AI / Stanford).

You finished Supervised Machine Learning in Course 1. This track picks up with neural networks (also called deep learning), then practical system-building advice, and finally decision trees.

Notes

These pages follow the Coursera videos and labs, rewritten for high school readers with review questions and plots where they help.

Anatomy of representative neurons in the human brain, with dendrites, cell body, nucleus, and axon labeled

Neural Networks Intuition
machine-learning
neural-networks

Course 2 Week 1 in one page: welcome, biological motivation, demand prediction, layers, activations, multi-layer architecture, and face recognition intuition (formal model…

Jul 9, 2026

Neural Network Model
machine-learning
neural-networks

Forward pass math for a four-layer neural network, covering the general activation formula, the layer-index superscript notation, and the input-as-layer-zero convention.

Jul 10, 2026

Forward Propagation
machine-learning
neural-networks

The forward propagation algorithm for neural network inference, worked through a handwritten digit recognition example with a 64-25-15-1 network.

Jul 10, 2026

TensorFlow Implementation
machine-learning
neural-networks

Forward propagation as real TensorFlow code, plus how NumPy matrices, row and column vectors, and TensorFlow tensors represent the same data differently.

Jul 10, 2026

Neural Networks in NumPy
machine-learning
neural-networks

Forward propagation from scratch in NumPy, first hardcoded neuron by neuron, then generalized into a single reusable dense layer function.

Jul 11, 2026

Lab: Handwritten Digit Recognition
machine-learning
neural-networks

A real neural network trained on 1000 real handwritten digit images to tell zeros from ones, built with the same Sequential pattern from the previous pages.

Jul 11, 2026

Training Neural Networks
machine-learning
neural-networks

The three-step TensorFlow pattern for training a neural network, specify the model, compile it with a loss function, and fit it to data.

Jul 12, 2026

Activation Functions
machine-learning
neural-networks

How ReLU, linear, and sigmoid activations work, and how to choose them for the output layer and the hidden layers of a neural network.

Jul 13, 2026

Multiclass Classification
machine-learning
neural-networks

Classification problems where the label can take more than two values, such as recognizing all ten digits, diagnosing several diseases, or spotting different defect types.

Jul 14, 2026

Additional Neural Network Concepts
machine-learning
neural-networks

The Adam optimization algorithm for faster training, and the convolutional layer, where each neuron reads only a window of the input.

Jul 14, 2026

Lab: Handwritten Digit Recognition (Multiclass)
machine-learning
neural-networks

A softmax neural network trained on 5000 real handwritten digit images to recognize all ten digits, 0 through 9, with a prediction grid and a look at its mistakes.

Jul 14, 2026

Model Evaluation and Selection
machine-learning
neural-networks

How to evaluate a trained model with a train/test split, and how to choose between models with a cross-validation set without fooling yourself.

Jul 14, 2026

Bias and Variance
machine-learning
neural-networks

Diagnosing underfitting and overfitting with J_train and J_cv, regularization and lambda, baselines, learning curves, the six fixes, and bias and variance in neural networks.

Jul 15, 2026

Machine Learning Development Process
machine-learning
neural-networks

The iterative loop of machine learning development, illustrated by building an email spam classifier. Choose an architecture, train, run diagnostics such as error analysis…

Jul 15, 2026

Skewed Datasets: Precision, Recall, and the F1 Score
machine-learning
neural-networks

Why accuracy misleads on rare classes, and how the confusion matrix, precision, recall, and the F1 score evaluate classifiers on skewed data.

Jul 16, 2026

Lab: Advice for Applying Machine Learning
machine-learning
neural-networks

Hands-on lab tuning polynomial degree, regularization, and data size with train, cross-validation, and test splits, then diagnosing a neural network.

Jul 16, 2026

Decision Trees
machine-learning
neural-networks

What a decision tree is, nodes and predictions, many possible trees, the step-by-step building process, and the criteria for when to stop splitting.

Jul 16, 2026

Decision Tree Learning
machine-learning
neural-networks

Entropy, information gain, the recursive tree-building algorithm, one-hot encoding, continuous features, and regression trees that predict numbers.

Jul 17, 2026

Tree Ensembles
machine-learning
neural-networks

Tree ensembles from sampling with replacement to random forests and XGBoost, plus when to choose decision trees versus neural networks.

Jul 17, 2026

Lab: Decision Trees
machine-learning
neural-networks

Practice lab implementing a decision tree from scratch, entropy, dataset splitting, information gain, and recursive building, to classify mushrooms as edible or poisonous.

Jul 17, 2026

Support Vector Machines
machine-learning
classification
support-vector-machines

The widest-street idea behind support vector machines, which training points the boundary depends on, what C and the kernel trick do, and where SVMs still win.

Sep 1, 2026
No matching items
Back to top