0.25
0.5
0.75
1.25
1.5
1.75
2
Lock-Free Approaches to Parallelizing Stochastic Gradient Descent
Published on Jan 25, 20127796 Views
Stochastic Gradient Descent (SGD) is a very popular optimization algorithm for solving data-driven machine learning problems. SGD is well suited to processing large amounts of data due to its robustne
Related categories
Chapter list
Lock-Free Approaches to Parallelizing Stochastic Gradient Descent00:00
Incremental Gradient Descent00:27
Example: Computing the mean - 0103:23
Example: Computing the mean - 0204:04
Convergence Rates - 0105:28
Convergence Rates - 0206:48
SGD and BIG Data07:37
Is SGD inherently Serial?08:57
HOGWILD!13:23
“Sparse” Function:14:50
Sparse Support Vector Machines17:20
Matrix Completion18:50
Graph Cuts20:34
Convergence Theory21:53
Hogs gone wild!24:35
Speedups26:49
JELLYFISH - 0129:14
JELLYFISH - 0232:19
JELLYFISH - 0332:53
Example Optimization - 0135:04
Example Optimization - 0235:34
Example Optimization - 0337:05
Simplest Problem? Least Squares39:20
Simple Question40:59
What about generically? - 0142:50
What about generically? - 0242:56
What about generically? - 0343:58
But what about on average?46:40
Summary48:59