What is the Optimal Number of Features? A learning theoretic perspective thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

What is the Optimal Number of Features? A learning theoretic perspective

Published on Feb 25, 20076904 Views

In this paper we discuss the problem of feature selection for supervised learning from the standpoint of statistical machine learning. We inquire what subset of features will lead to the best classifi

Related categories

Chapter list

What is The Optimal Number of Features? A learning theoretic Perspective 00:01
What is Feature Selection?00:18
Reasons to do Feature Selection01:09
The Questions02:10
Two Gaussians - Problem Setting02:42
Problem Setting – Cont.04:18
Illustration05:42
Result07:23
Solving for Specific 08:07
Solving for Specific  - Cont.10:33
Problem Setting – Cont.11:40
Solving for Specific  - Cont.11:59
Proof 12:13
Proof 14:24
Proof – Cont.14:49
Proof – Cont.15:53
“Empirical Proof” of the Lemma16:37
Linear SVM Error (averaged on 200 repeats, c=0.01, using Gavin Cawley’s tool box)17:25
Conclusions19:34
What is The Optimal Number of Features? A learning theoretic Perspective 21:09
Proof – Cont.24:30