Fast Food: Approximating Kernel Expansion in Loglinear Time thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Fast Food: Approximating Kernel Expansion in Loglinear Time

Published on Jan 18, 20139450 Views

The ability to evaluate nonlinear function classes rapidly is crucial for nonparametric estimation. We propose an improvement to random kitchen sinks that offers O(n log d) computation and O(n) st

Related categories

Chapter list

Fastfood: O(n log d) feature maps for kernels00:00
The trouble with kernels - 100:01
The trouble with kernels - 203:39
Random Kitchen Sinks04:40
Key Idea06:45
Properties09:27
Matrix approximation error11:00
Generalization Performance11:33
Speed & accuracy11:48
Summary13:01