IDS 705: Principles of Machine Learning

Schedule and Syllabus

The schedule below is a guide to what we will be covering throughout the semester and is subject to change to meet the learning goals of the class. Check this website regularly for the latest schedule and for course materials that will be posted here through links on the syllabus.
ISL = Introduction to Statistical Learning, by James, Witten, Hastie, and Tibshirani
DM = Introduction to Data Mining, by Tan, Steinbach, Karpatne, and Kumar
PRML = Pattern Recognition and Machine Learning, by Bishop
DL = Deep Learning, by Goodfellow, Bengio, and Courville
RL = Reinforcement Learning: An Introduction, by Sutton and Barto

Event Type	Date	Description	Readings	Course Materials
Lecture 1	Wednesday Jan 20	What is machine learning? Course overview and an orientation to the major branches of machine learning: supervised, unsupervised, and reinforcement learning	None	[slides]
		Module 1: Supervised Learning
Lecture 2	Monday Jan 25	An end-to-end machine learning example Stating the problem, creating the model, evaluating performance, and operationalizing the solution.	ISL Ch. 1 + 2.1	[slides] [sample code]
Lecture 3	Wednesday Jan 27	How flexible should my algorithms be: the bias-variance tradeoff K-nearest neighbors classification and the bias-variance tradeoff	ISL 2.2	[slides]
Lecture 4	Monday Feb 1	Linear Models I Simple linear regression, multiple linear regression, measuring error, model fitting and least squares, comparing linear regression and classification	ISL Intro of 3, 3.1, and 3.2	[slides]
Deliverable	Tuesday Feb 2	Assignment #1 Due (at 9pm)		[assignment] [sample 13] [submit]
Lecture 5	Wednesday Feb 3	Linear Models II Nonlinear transformations of predictors, cost/loss functions, selecting parameters through gradient descent.	ISL 3.3 and 3.5	[slides]
Lecture 6	Monday Feb 8	Performance evaluation and model comparison Choosing the right model: accuracy vs speed vs interpretability; metrics for supervised learning performance evaluation: types of errors, receiver operating characteristics curves, and confusion matrices	ISL 4.1, 4.2, and 4.3	[slides]
Lecture 7	Wednesday Feb 10	Resampling methods for performance evaluation: model validation and testing strategies How to use model performance metrics to measure metrics of generalization performance; resampling techniques: training, testing, and validation datasets and cross validation; common pitfalls around biased sampling and data snooping/leakage	ISL 5.1 and 5.2	[slides]
Lecture 8	Monday Feb 15	Decision theory A risk-based framework for determining to operate supervised learning algorithms in practice; choosing ROC operating points through risk-minimization and how application-specific costs associated with different types of errors can be used to determine optimal operating points for classifiers	Link to reading	[slides]
Deliverable	Tuesday Feb 16	Assignment #2 Due (at 9pm)		[assignment] [submit]
Lecture 9	Wednesday Feb 17	Reducing overfit Feature selection; Occam’s razor; Subset selection; L1 (ridge), L2 (LASSO), and elastic net regularization; early stopping.	ISL 6.1 and 6.2	[slides]
Lecture 10	Monday Feb 22	Generative models for classification Generative vs discriminative models; linear discriminant analysis, quadratic discriminant analysis, and naïve Bayes	ISL 4.4 and 4.5	[slides]
Deliverable	Tuesday Feb 23	Project Proposal Due (at 9pm)		[project] [submit]
Lecture 11	Wednesday Feb 24	Tree-based models and ensembles From decision trees to random forests: bagging, bootstrapping, and boosting	ISL 8.1 and 8.2	[slides]
Lecture 12	Monday Mar 1	Kernel Methods Introducing Kernel machines via the kernel perceptron, maximum margin classifiers, and support vector machines	ISL Ch 9: 9.1-9.4	[slides]
Lecture 13	Wednesday Mar 3	Neural networks I Introduction to neural networks and representation learning; forward propagation, network architecture, and how to adapt to regression or classification problems	PRML Ch 5: 5.1	[slides]
Deliverable	Tuesday Thursday Mar 4	Assignment #3 Due (at 9pm)		[assignment] [submit]
Lecture 14	Monday Mar 8	Neural networks II Fitting a neural network to training data through gradient descent and backpropagation; how backpropagation is used to compute gradients in neural networks; hyperparameters and architecture choices in neural networks and practices for training neural networks successfully	PRML Ch 5: 5.3 (intro), 5.3.1, 5.3.2, and Calculus on Computational Graphs	[slides]
No Class	Wednesday Mar 10
Lecture 15	Monday Mar 15	Introduction to Deep learning Common architectures of deep learning models, in particular convolutional neural networks for computer vision and the tools used to implement them	Optional Quiz DL Ch 11: Practical Methodology	[slides]
		Module 2: Unsupervised Learning
Lecture 16	Wednesday Mar 17	Dimensionality reduction The Curse of Dimensionality and intro to principal components analysis (PCA)	ISL 6.3, 6.4, 10.1, and 10.2	[slides]
Deliverable	Thursday Mar 18	Assignment #4 Due (at 9pm)		[assignment] [submit] [NN Math Guide]
Lecture 17	Monday Mar 22	Principal components analysis (PCA) Explaining how PCA works and how we calculate the principal components.	ISL 10.3	[slides]
Deliverable	Tuesday Mar 23	Project Progress Report Due (at 9pm)		[project] [submit]
Lecture 18	Wednesday Mar 24	Clustering I From K-means to Gaussian mixture model clustering and Expectation Maximization	DM Ch 7 (link): Intro, 7.1 and 7.2	[slides]
Lecture 19	Monday Mar 29	Clustering II Hierarchical clustering, DBSCAN, and spectral clustering	DM Ch 7 (link): 7.3 and 7.4	[slides]
		Module 3: Reinforcement Learning
Lecture 20	Wednesday Mar 31	Reinforcement Learning I Formulating the reinforcement learning problem	RL Ch 1: 1.1-1.6; Ch 2: 2.1-2.5	[slides]
Lecture 21	Monday Apr 5	Reinforcement Learning II Policy and value functions, rewards, and introduction to Markov processes	RL Ch 3	[slides]
Deliverable	Saturday Monday Apr 5	Assignment #5 Due (at 9pm)		[assignment] [submit]
Lecture 22	Wednesday Apr 7	Reinforcement Learning III From Markov Chains to Markov Decision Processes (MDPs)	RL Ch 4	[slides]
No Class	Monday Apr 12
Lecture 23	Wednesday Apr 14	Reinforcement Learning IV Finding optimal policies through policy iteration, value iteration, and Monte Carlo methods	RL Ch 5: 5.1-5.3	[slides]
Lecture 24	Monday Apr 19	State-of-the-art machine learning and applications Cutting-edge applications and techniques: ideas on where the field is heading and how to stay up-to-date	None	[slides]
Deliverable	Wednesday Apr 21	Final project showcase (Project Video Due) (last class meeting of the semester)		[project]
Deliverable	Monday Thursday Apr 22	Assignment #6 Due (at 9pm)		[assignment] [submit]
Deliverable	Monday Apr 26	Final Project Report Due (at 9pm)		[project] [submit]
Deliverable	Monday Apr 26	Final Project Peer Evaluation Due (at 9pm)		[project] [submit via emailed link]