Neural Information Processing Systems - NIPS05 Workshops

Large-scale parallel implementations of SVMs

author: Igor Durdanović, NEC Laboratories America, Inc.

Categories

Top: Computer Science: Machine Learning: Kernel Methods: Support Vector Machines

You might be experiencing some problems with Your Video player.

*Slides*
0:00	Large-scale parallel SVM implementation
0:15	Related Work
1:03	The Problem: Regular, plain SVM
1:12	The Algorithm
1:55	The Algorithm1
2:56	Engineering
3:36	Engineering in practice
4:51	Vector-type optimized Kernels
6:59	Sorting the data set by labels
7:49	Multi-threading on multi-processor machines
8:36	Paralelization: Spread-Kernel: full data [2 nodes]
10:17	Paralelization: Spread-Kernel: full data [p nodes]
11:05	The network max( WorkingSet ) [p nodes]
12:15	Paralelization: Spread-Kernel: split data [2 nodes]
13:16	Paralelization: Spread-Kernel: split data [p nodes]
13:30	Reliable MULTICAST
14:15	NEC Cluster
14:40	Results: Speedup: theoretical model
15:59	Results: Speedup: Training Forest [522K samples]
16:45	Results: Speedup: Training MNIST [220K samples]
17:03	Results: Speedup: Training MNIST [500K samples]
17:15	Results: Speedup: Training MNIST [1M samples]
17:26	Results: Speedup: Training MNIST [2M samples]
17:30	Results: Speedup: Training MNIST [4M samples]
17:36	Summary
17:56	Software [availability to be determined]

Lecture rating

People found this lecture:
Worth seeing
because it is:
Valuable and informative
Well presented
Easily understandable
Acceptably recorded
You need to login to cast your vote.

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.