Difference between revisions of "10-601 Matrix Factorization"

From Cohen Courses
Jump to navigationJump to search
Line 15: Line 15:
  
 
You should know:
 
You should know:
−
* What loss function and constraints are associated with PCA - i.e., what the "PCA optimization problem" is.
+
* What social recommendations systems are, and how they relate to matrix factorization.
−
* How to interpret the low-dimensional embedding of instances, and the "prototypes" produced by PCA and MF techniques.
+
* How to solve MF via gradient descent.
−
* How to interpret the prototypes in the case of dimension reduction for images.
+
* How matrix factorization is related to PCA and k-means.

Revision as of 12:39, 13 April 2016

This a lecture used in the Syllabus for Machine Learning 10-601B in Spring 2016

Slides

Readings

Matrix factorization and collaborative filtering is not covered in Murphy or Mitchell. Some external readings are below.

  • Koren, Yehuda, Robert Bell, and Chris Volinsky. "Matrix factorization techniques for recommender systems." Computer 8 (2009): 30-37.
  • There's a nice description of the gradient-based approach to MF, and a scheme for parallelizing it,by Gemulla et al.

Summary

You should know:

  • What social recommendations systems are, and how they relate to matrix factorization.
  • How to solve MF via gradient descent.
  • How matrix factorization is related to PCA and k-means.
Retrieved from "http://curtis.ml.cmu.edu/w/courses/index.php?title=10-601_Matrix_Factorization&oldid=18482"