Supervised Machine Learning for Engineers

This module provides a foundational overview of machine learning concepts and frameworks. Students learn about the basic terminology, core challenges in machine learning, and different learning paradigms including supervised, semi-supervised, unsupervised, and reinforcement learning

This module covers fundamental regression techniques for numerical prediction. Students learn about different error metrics, handling outliers, and implementing basic regression models. The focus is on practical applications of linear regression and K-nearest neighbor regression

This module explores classification techniques for categorical prediction. Students learn about various classifier types, including histogram-based methods, nearest mean classifiers, and K-nearest neighbor classifiers. The module also covers theoretical concepts like the Bayes classifier and Bayes error

This module focuses on gradient descent and its variants for model training. Students learn about batch, mini-batch, and stochastic gradient descent methods, along with their practical applications in training logistic regression models

This module addresses the critical challenge of overfitting in machine learning. Students explore using linear models for nonlinear tasks, understanding the bias-variance trade-off, and learning about the curse of dimensionality. The module also covers techniques for estimating data requirements

This module covers techniques for model validation and regularization. Students learn about cross-validation methods, hyperparameter tuning, and regularization techniques including Ridge regression and LASSO, with a focus on preventing overfitting

This module focuses on methods for evaluating classifier performance. Students learn about baseline accuracy metrics, confusion matrices, and ROC curves, enabling them to properly assess and compare different classification models

This module covers the theory and application of Support Vector Machines. Students learn about margin concepts, hard-margin and soft-margin SVMs, and kernel methods for handling non-linear classification problems

This module explores decision tree algorithms and their applications. Students learn about the CART algorithm for training decision trees, understanding their interpretability advantages, and managing their tendency to overfit

This module involves applying learned concepts to build a complete machine learning pipeline for handwritten digit recognition using the MNIST dataset. Students implement hyperparameter tuning and evaluation techniques in a practical setting