Location: Conferences » Other » The 14th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

Regularization Paths and Coordinate Descent

author:Trevor Hastie, Stanford University
published: Sept. 26, 2008,   recorded: August 2008,   views: 511
You might be experiencing some problems with Your Video player.

Slides

Slides
0:00 - Introduction
4:23 Fast Regularization Paths via Coordinate Descent
5:36 Linear Models in Data Mining
9:09 The elements of statistical learning - the book
10:13 Linear regression via the Lasso (Tibshirani, 1995)
12:34 Brief History of `1 Regularization
14:50 Lasso Coefficient Pro le
16:55 History of Path Algorithms
17:10 Lasso Coefficient Pro le
17:22 History of Path Algorithms
21:01 Coordinate Descent
22:33 Graph - LARS and GLMNET
23:06 Speed Trials
24:39 Linear Regression | Dense Features
27:37 Logistic Regression | Dense Features
28:17 Logistic Regression | Sparse Features
29:20 Logistic Regression | Real Datasets
32:12 A brief history of coordinate descent for the lasso (1)
32:40 A brief history of coordinate descent for the lasso (2)
32:59 A brief history of coordinate descent for the lasso (3)
33:55 Coordinate descent for the lasso
35:34 Why is coordinate descent so fast?
38:43 Why is coordinate descent so fast? (2)
39:57 Binary Logistic Models
40:59 Elastic-net Penalty
43:18 Leukemia Data, Logistic, N=72, p=3571, rst 10 steps shown
44:20 Multiclass classi cation
45:03 Summary
46:33 Other Applications
48:43 Undirected Graphical Models
49:07 Grouped lasso
50:23 CGH modeling and the fused lasso.
51:17 Summary

Related content

Visitors who watched this lecture also watched...

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
Lecture popularity: You need to login to cast your vote.

Description

In a statistical world faced with an explosion of data, regularization has become an important ingredient. In many problems, we have many more variables than observations, and the lasso penalty and its hybrids have become increasingly useful. This talk presents some effective algorithms based on coordinate descent for fitting large scale regularization paths for a variety of problems. Joint work with Rob Tibshirani and Jerome Friedman

See Also:

Download slides icon Download slides: kdd08_hastie_rpcd_01.pdf (670.7 KB)

Launch Windows Media PlayerLaunch in a standalone WM Player

WMedia PlayerSwitch to Windows Media Player

View slides View slides

Help icon Streaming Video Help

WebLink icon Windows Media Player Firefox Plugin - Download

Link this page  

Would you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !

Write your own review or comment:

make sure you have javascript enabled or clear this field: