Sign inSign up
Online PCA with Spectral Bounds thumbnail
slide-image
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Online PCA with Spectral Bounds

Published on Aug 20, 20152258 Views

Edo Liberty

This paper revisits the online PCA problem. Given a stream of $n$ vectors $x_t \in \R^d$ (columns of $X$) the algorithm must output $y_t \in \R^\ell$ (columns of $Y$) before receiving $x_{t+1}$. The g

COLT 2015 - Paris

Related categories

Active LearningComputational Learning TheoryOn-line LearningReinforcement LearningSemi-supervised Learning

Chapter list

Yahoo! Online PCA with Spectral Bounds00:00
Principal Component Analysis00:12
Optimal Offline Solution00:38
Online PCA01:16
Online PCA, Possible Problem Definitions01:31
Noisy Data Spectra01:49
Online PCA Problem Definitions02:32
Algorithm Intuition - 103:12
Algorithm Intuition - 203:18
Algorithm Intuition - 303:22
Algorithm Intuition - 403:25
Algorithm Intuition - 503:36
Algorithm Properties03:47
Fixing the Algorithm04:28
Visual Ilustration and Open Problem05:09
Thank you05:18