0.25
0.5
0.75
1.25
1.5
1.75
2
Statistical Translation, Heat Kernels, and Expected Distances
Published on Feb 25, 20074670 Views
High dimensional structured data such as text and images is often poorly understood and misrepresented in statistical modeling. The standard histogram representation suffers from high variance and p
Related categories
Chapter list
Statistical Translation, Heat Kernels, and Expected Distances00:00
Motivation00:07
Motivation0102:18
A related example: query expansion03:17
A related example: query expansion0104:06
Statistical translation for document modeling04:36
Statistical translation for document modeling0104:49
Interpretations of the model06:21
Interpretations of the model0107:33
Assumption about document translation08:09
Assumption about document translation0108:54
Estimating Tij = P (Wi --> Wj)09:31
Estimating Tij = P (Wi --> Wj)0110:04
Estimating Tij = P (Wi --> Wj)0210:35
Estimating Tij = P (Wi --> Wj)0311:02
Estimating Tij = P (Wi --> Wj)0411:28
Word translation result12:10
Expected Distance13:07
Expected Distance0114:42
RCV1 Document classification results15:11
RCV1 Document classification results0116:26
Conclusion16:42