The sample complexity of agnostic learning under deterministic labels thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

The sample complexity of agnostic learning under deterministic labels

Published on Jul 15, 20143250 Views

With the emergence of Machine Learning tools that allow handling data with a huge number of features, it becomes reasonable to assume that, over the full set of features, the true labeling is (almost)

Related categories

Chapter list

The sample complexity of agnostic learning under deterministic labels00:00
A non-investigated corner in (agnostic) PAC theory00:10
Why care?01:45
Results: Noise in labels is not the only source of diculty02:40
Results: Learning rates vary with the class03:15
Results: ERM not always optimal04:13
Results: Unlabeled data is useful04:54
Summary05:30
See you at the poster!06:11