Difference between revisions of "Syllabus for Machine Learning with Large Datasets 10-605 in Spring 2014"

Revision as of 16:16, 24 March 2014

Notes:

The assignments are from 2013, and will be modified over the course of the semester - some may be changed substantially.
Lecture notes will be posted around the time of the lectures.

Mon Jan 13. Overview of course, cost of various operations, asymptotic analysis.
Wed Jan 15. Review of probabilities, joint distributions and naive Bayes
Mon Jan 20. No class - Martin Luther King Day.
Wed Jan 22. Streaming algorithms and Naive Bayes; The stream-and-sort design pattern; Naive Bayes for large feature sets.
- New Assignment: streaming Naive Bayes 1 (with feature counts in memory). PDF Handout
Mon Jan 27. Messages and records 1; Phrase finding.
- Assignment due: streaming Naive Bayes 1 (with feature counts in memory).
Wed Jan 29. Phrase Finding and Rocchio
- New Assignment: streaming Naive Bayes 2 (with feature counts on disk) with stream-and-sort. PDF Handout
Thursday Jan 30. Scheduled down-time for the wiki host. (Obviously, it's up again now!)

Mon Feb 3. Rocchio and Parallel Perceptrons
Wed Feb 5. Perceptrons/Map-reduce and Hadoop.
- Assignment due: streaming Naive Bayes 2 (with feature counts on disk) with stream-and-sort
- New Assignment: phrase finding with stream-and-sort. PDF Handout
Mon Feb 10. Parallel Perceptrons.
Wed Feb 12. Guest lecture: Matt Hurst, Microsoft/Bing: Local Search at Bing. One-on-one meetings with Matt can be scheduled for Thursday 12/13 between 9-12 in Gates-Hillman 6501, afternoon meetings 12:30-1:30pm in Gates-Hillman 6002.
Mon Feb 17. Scalable SGD and Hash Kernels
- Assignment due: phrase finding with stream-and-sort
- New Assignments: Naive Bayes with Streaming Hadoop, Naive Bayes with Streaming Hadoop & Phrase-finding with Hadoop. PDF Handout (4a)PDF Handout (4b)PDF Handout (4c)
Wed Feb 19. Matrix Factorization and SGD, plus another Hadoop demo
Fri Feb 21. Nothing due - the streaming run for Naive Bayes, 4(a), has been postponed till Monday.
Mon Feb 24. SGD for Matrix Factorization, and Randomized Algorithms 1 (Bloom Filters)
- Streaming run on Hadoop of Naive Bayes due
Wed Feb 26. Randomized Algorithms
Fri Feb 28.
- Non-streaming run on Hadoop of Naive Bayes due.

Mon Mar 3. Guest Lecture: Garth Gibson, Cloud Computing and Programming Paradigms
Wed Mar 5. Guest lecture: Alex Beutel, SGD on Hadoop
Fri Mar 7.
- Hadoop assignment (phrase-finding) due
Mon Mar 10. no class - spring break.
Wed Mar 12. no class - spring break.
Mon Mar 17. Scalable PageRank
- New Assignment: memory-efficient SGD PDF handout
Wed Mar 19. Subsampling a graph with RWR
Mon Mar 24. Subsamping continued and SSL on Graphs
Wed Mar 26. Understanding spectral clustering techniques.
- Assignment due: memory-efficient SGD
Mon Mar 31. Sparse sampling and parallelization for LDA
- New Assignment: Subsampling and visualizing a graph. PDF writeup

Wed Apr 2. Speeding up LDA-like models: All-reduce and online LDA
- Assignment due: Subsampling and visualizing a graph.
- New Assignment: Workflows with Pig
Mon Apr 7. Fast KNN and similarity joins 1.
Wed Apr 9. Fast KNN and similarity joins 2.
Mon Apr 14. Scaling up decision tree learning
Wed Apr 16. Gradient boosting with trees
- Assignment due: Workflows with Pig
Mon Apr 21. TBA
Wed Apr 23. TBA
Mon Apr 28. Exam review session.
Wed Apr 30. In-class exam.

@@ Line 51: / Line 51: @@
 * Wed Mar 26. [[Class meeting for 10-605 Spectral Clustering|Understanding spectral clustering techniques.]]
 ** '''Assignment due: memory-efficient SGD'''
+* Mon Mar 31. [[Class meeting for 10-605 LDA 1|Sparse sampling and parallelization for LDA]]
 ** ''New Assignment: Subsampling and visualizing a graph.'' [http://www.cs.cmu.edu/~wcohen/10-605/assignments/snowball.pdf PDF writeup]
-* Mon Mar 31. [[Class meeting for 10-605 LDA 1|Sparse sampling and parallelization for LDA]]
 == April and May ==