en
0.25
0.5
0.75
1.25
1.5
1.75
2
Probabilistic Interpretation of Quasi-Newton Methods
Published on Jan 15, 20133373 Views
This talk is a case-study about the utility of probabilistic formulations for numerical mathematics. I present a recent result showing that quasi-Newton methods can be interpreted as performing Gaussi
Related categories
Chapter list
Probabilistic Interpretation of Quasi-Newton Methods00:00
Optimization is Inference00:33
Quasi-Newton methods04:33
QN methods and Gaussian regression08:33
Benefit # 109:55
Encoding Symmetry (1)10:27
Encoding Symmetry (2)12:58
What does this mean?14:39
Benefit # 216:31
Nonparametric Quasi-Newton methods19:02
. . . and immediate benefits (1)23:05
. . . and immediate benefits (2)24:33
. . . and immediate benefits (3)25:50
. . . and immediate benefits (4)28:39
Benefit # 328:57
Convergence Properties29:21
Challenge # 131:20
Summary32:09