Lecture Topic | Readings | Assignments | ||
Aug | 25 | Intro: what is machine learning? Basic concepts of supervised learning with examples | Review material for Week 1 discussion section, Readings for Week 1 | A0: Linear algebra, optimization, probability (due 9/2) |
Aug | 30 | Linear regression, least squares | Readings for Week 2 | |
Sep | 1 | Linear regression contd.; application (global seismic monitoring) | A1: Predicting travel times for seismic waves through the Earth (due 9/9) | |
6 | Machine learning methodology: learning curves, overfitting, regularization, cross-validation, feature selection | Readings for Week 3 | ||
8 | Classification, 0/1 loss, linear classifiers, SVMs | A2: Classification of seismic wave types using SVMs (due 9/19) | ||
13 | Logistic regression | Readings for Week 4 | ||
15 | Kernelization of SVMs and other models | |||
20 | Decision tree learning | Readings for Week 5 | A3: Decision tree and ensemble learning applied to seismic phase classification (due 10/3) | |
22 | Ensemble learning methods (bagging, boosting, etc.) | |||
27 | Instance-based methods (k-nearest-neighbor, interpolation, etc.) | Readings for Week 6 | ||
29 | Multilayer perceptrons ("neural networks"), gradient-based optimization, applications | |||
Oct | 4 | Instance-based learning contd.: distance metrics and efficient indexing | Readings for Week 7 | |
6 | Midterm | |||
11 | Theoretical analysis: generalization error bounds, regret bounds for online learning | Readings for Week 8 | ||
13 | Probabilistic methods: ML, MAP, Bayesian learning, naive Bayes, "bag-of-" models | A4: Spam filtering with Naive Bayes (due 10/21) | ||
18 | Gaussian discriminants; logistic regression revisited | Readings for Week 9 | ||
20 | Special presentation: Big data in Groupon | A5: Credit scoring with Gaussian discriminants and logistic regression (due 10/28) | ||
25 | Bayesian regression | Readings for Week 10 | ||
27 | Density estimation: kernel density estimation, mixture models | A6: Estimating earthquake probabilities (due 11/9) | ||
Nov | 1 | K-means, EM | Readings for Week 11 | |
3 | Bayes nets: representation | |||
8 | Bayes nets: inference, learning | Readings for Week 12 | ||
10 | Bayes net learning contd. | A7: Bayes net for car insurance (due 11/20) | ||
15 | Time series models (Markov processes, n-grams, AR models) | Readings for Week 13 | ||
17 | Time series models contd. (HMMs, dynamic Bayes nets) | |||
22 | Learning in computer vision | Readings for Week 14 | ||
24 | Thanksgiving | |||
29 | Sequential analysis, bandits, active learning | Readings for Week 15 | ||
Dec | 1 | Summary, current and future developments | ||
6 | Reading/Review/Recitation | |||
8 | Reading/Review/Recitation | |||
16 | Final (7pm - 10pm) | Location TBD |