Theoretical foundations, algorithms, methodologies, and applications for machine learning. Topics may include supervised methods for regression and classication (linear models, trees, neural networks, ensemble methods, instance-based methods); generative and discriminative probabilistic models; Bayesian parametric learning; density estimation and clustering; Bayesian networks; time series models; dimensionality reduction; programming projects covering a variety of real-world applications.
Syllabus: [pdf]Week | Topics | Slides | Reading | Discussion | Homework |
1-2 | Introduction, MLE | [pdf] | Murphy 2.1-2.5 | [pdf] | |
3 | Decision theory, linear algebra review, Gaussians | [pdf] [pdf] | Murphy 4.1-4.2 (starred sections optional) Multivariate Gaussians LDA |
[pdf] | HW2 solutions [pdf] |
4 | Linear regression | [pdf] | Murphy 7.1-7.3, 7.5.1 | [pdf] | |
5 | Logistic regression, optimization, support vector machines | [pdf] | Murphy 8.1-8.3.3 | [pdf] | |
6-7 | SVMs continued, nonparametric methods | [pptx] |
Murphy 14.5 (excluding 14.5.1) Burges [pdf] Loss Functions [pdf] | [pdf] | |
8 | Midterm | [pdf] | |||
9 | Nearest neighbor | [pdf] | Benefits of Non-Parametric Methods [pdf] (only read introductory part) | [pdf] | |
10 | Decision trees | [ppt] | Adaboost [pdf] | [pdf] | |
11 | Neural networks | [ppt] | Backpropagation [pdf] Lectures from Professor Yaser Abu-Mostafa (Caltech) |
[pdf] | |
12 | Unsupervised learning, clustering | [pdf] | [pdf] | ||
13 | Mode seeking | [ppt] | [pdf] | ||
14 | Dimensionality reduction, PCA | [ppt] | SVD [pdf] | [pdf] |
Prof. Jitendra Malik
malik@eecs.berkeley.edu
Office Hours: Monday 11-12, 722 Sutardja Dai
Prof. Alyosha Efros
efros@eecs.berkeley.edu
Office Hours: Tuesday 3:40-5, 724 Sutardja Dai
Jonathan Ho
jonathanho@berkeley.edu
Office Hours: Thursday 11-12pm, 651 Soda
Sharad Vikram
sharad.vikram@gmail.com
Office Hours: Friday 12-1pm, 611 Soda
Christopher Xie
chrisdxie@berkeley.edu
Office Hours: Tuesday 4-5pm, 411 Soda
Ning Zhang
nzhang@eecs.berkeley.edu
Office Hours: Monday 10-11am, 611 Soda