2017/1
Department of Computing
This schedule is tentative and subject to change. Please check back often.
The first lecture is Tuesday, March 7th, 2016. That is, we start during the second week of classes.
| Date | Lecture | Topics | Readings | Announcements |
|---|---|---|---|---|
Introduction and Background |
||||
| Tue | Lecture:
Introductory class |
Discussion of machine learning in general, including context, motivation, and main tasks. We also cover the logistics of the course. |
|
|
Machine Learning Foundations |
||||
| Thu | Lecture:
Regression |
Gradient descent, estimating parameters through closed form, interpretation of estimated model parameters, prediction using estimated model, polinomial regression |
|
|
| Tue | Lecture:
Regression Lab. |
Predicting house prices. |
|
|
| Thu | Lecture:
Classification |
Linear classifiers, logistic regression, decision boundaries, degree of confidence in prediction, class probabilities interpretation, 1-hot encoding, multiclass classification. |
|
|
| Tue | Lecture:
Classification Lab. |
Sentiment analysis. |
|
|
| Thu | Lecture:
Clustering |
Nearest neighbor search, document representations, tf-idf, measuring document similarity, KD-trees, locality sensitive hashing. |
|
|
| Tue, 4/4/17 | Lecture:
Clustering Lab. |
Document retrieval. |
|
HW1 Due |
Evaluation |
||||
| Thu, 4/6/17 | Lecture:
Model Assessment and Selection |
Loss function definition, training-generalization-test error, tradeoffs in training/test splits, irreducible error, bias, variance, parameter tuning. Precision, recall, trade-off precision/recall, precision/recall curves. |
|
|
| Tue, 4/11/17 | Lecture:
Project Interviews |
A short description of the project, including: a title, name of the members, a short description of your proposed project, datasets, and key references (at least 3 refs). It is expected about a page. |
|
Project Proposal Due |
| Thu, 4/13/17 | Lecture (No class):
Holiday |
|
|
|
| Tue, 4/18/17 | Lecture:
|
Midterm |
|
Midterm |
Recommender Systems |
||||
| Thu, 4/20/17 | Lecture:
Recommender Systems [Slides] |
Content-based, collaborative filtering, hybrid recommendation, evaluation, top-N recommendation, cold-start, research challenges. |
|
|
| Tue, 4/25/17 | Lecture:
Recommender Systems Lab. |
Product recommendation. |
|
|
Regression |
||||
| Thu, 4/27/17 | Lecture: Ridge regression: regulating overfitting when using many features | Ridge regression cost function, tuning learning rate, estimation of parameters (in closed form and using iterative gradient descent algorithm), k-fold cross validation to tune learning rate. |
|
|
| Tue, 5/2/17 | Lecture:
Project Practice Lab |
|
|
|
| Thu, 5/4/17 | Lecture:
Lasso regularization: regularization for feature selection |
Feature selection, all subsets, forward stepwise, lasso objective, lasso coefficient interpretation, L1 penalty. |
|
|
| Tue, 5/9/17 | Lecture:
Project Practice Lab |
|
|
|
Classification |
||||
| Thu, 5/11/17 | Lecture:
Linear classifiers: overfitting and regularization |
Overfitting, regularization, decision boundaries, L1 and L2 regularization, estimation of learning rate. |
|
|
| Tue, 5/16/17 | Lecture:
Project Practice Lab |
|
HW2 Due |
|
| Thu, 5/18/17 | Lecture:(No class)
|
|
|
|
| Tue, 5/23/17 | Lecture (No class):
|
|
|
|
| Thu, 5/25/17 | Lecture:
Decision trees |
Definition, output interpretation, greedy algorithm, traversing a decision tree to: majority class predictions, probability predictions, and multiclass predictions. Overfitting in decision trees, tree depth, node split, classification error, tree complexity. |
|
Midway Report Due |
| Tue, 5/30/17 | Lecture:
Project Practice lab. |
|
|
|
| Thu, 6/1/17 | Lecture:
Handling missing data and Ensembles |
Skipping and imputation of missing values, modification of learning algorithm. Ensembles formalization, boosting, AdaBoost, data weights, decision stumps. |
|
|
| Thu, 6/6/17 | Lecture:
SVM, Stochastic Gradient, and Online Learning |
Maximal margin classifier, support vector classifier, support vector machine. Speedup learning using sthocastic gradient, online learning problems, relating sthocastic gradient to online learning. |
|
|
Unsupervised learning |
||||
| Thu, 6/8/17 | Lecture:
Mixture models and EM algorithm |
Probabilistic model-based clustering using mixture models, soft assignments, EM algorithm. |
|
|
| Tue, 6/13/17 | Lecture:
Topic models and Hierarchical clustering |
Latent Dirichlet Allocation, contrast clustering and mixed membership models, bag-of-words representation, Gibbs sampling, collapsed Gibbs sampling for LDA. Hierarchical clustering. |
|
|
| Thu, 6/15/17 | Lecture (No class):
Holiday |
|
|
|
| Tue, 6/20/17 | Lecture:
Final exam |
|
Final exam |
|
Deep + Neural Networks |
||||
| Thu, 6/22/17 | Lecture:
Deep Learning |
Neural networks, Deep Learning, Convolutional Neural Networks, Recurrent Neural Networks, Autoenconders |
|
|
Presentations |
||||
| Tue, 6/27/17 | Lecture: |
Project Presentations |
|
Project Report Due |
| Thu, 6/29/17 | Lecture: |
Project Presentations |
|
|
Final Exams |
||||
| Tue, 7/4/17 | Lecture: |
Substitutive Exam |
|
|
| Thu, 7/6/17 | Lecture: |
E.E. |
|
|