Large Scale Machine Learning and Other Animals

Showing posts with label Lanczos. Show all posts
Showing posts with label Lanczos. Show all posts
Saturday, October 22, 2011

Speeding up SVD on a mega matrix using GraphLab - part 2

›
It seems I struck some nerve, since I am getting a lot of comments about this blog post. First, here is an email I got from Bryan , Resear...
Friday, October 21, 2011

Speeding up SVD computation on a mega matrix using GraphLab

›
About two weeks ago, I got a note from Tom Mitchell , head of the Machine Learning Dept. at Carnegie Mellon University, that he is looking f...
5 comments:
Monday, October 10, 2011

Lanczos Algorithm for SVD (Singular Value Decomposition) in GraphLab

›
A while ago, I examined Mahout's parallel machine learning toolbox mailing list, and found out that a majority of the questions where a...
2 comments:
Thursday, June 9, 2011

SVD (Lanczos algorithm) in GraphLab

›
I have now implemented the Lanczos algorithm in GraphLab, as part of GraphLab's collaborative filtering library . Here are some perf...
1 comment:
›
Home
View web version

About Me

My photo
Danny Bickson
6 years ago, along with my collaborators at Carnegie Mellon University, I have started the GraphLab large scale open source project, which is a framework for implementing machine learning algorithms in parallel and distributed settings. When the project became popular, we have decided to raise money to expand the project and provide an industry grade solution. Specifically I wrote the award wining collaborative filtering toolkit to GraphLab which is widely deployed today, and helped us win top places at ACM KDD CUP 2011, ACM KDD CUP 2012 among other competitions. Checkout our website: http://dato.com
View my complete profile
Powered by Blogger.