The Minimum Transfer Cost Principle for Model-Order Selection thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

The Minimum Transfer Cost Principle for Model-Order Selection

Published on Nov 30, 20112794 Views

The goal of model-order selection is to select a model variant that generalizes best from training data to unseen test data. In unsupervised learning without any labels, the computation of the general

Related categories

Chapter list

The Minimum Transfer Cost Principle for Model - Order Selection00:00
Outline00:24
Motivation00:48
Transfer Costs: Intuition02:01
Transfer costs for factorial models03:23
The Minimum Transfer Cost Principle05:25
Denoising Matrices via truncated SVD06:32
Selecting the rank for truncated SVD08:38
Correlation Clustering09:47
Finding the number of clusters12:16
Transfer costs for k-means13:51
k-means: easy case15:03
k-means: scale inhomogenity17:17
Conclusion18:58