The Rate of Convergence of AdaBoost thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

The Rate of Convergence of AdaBoost

Published on Aug 02, 20114674 Views

The AdaBoost algorithm was designed to combine many "weak" hypotheses that perform slightly better than random guessing into a "strong" hypothesis that has very low error. We study the rate at which

Related categories

Chapter list

The Rate of Convergence of AdaBoost00:00
AdaBoost (Freund and Schapire 97) - 100:16
AdaBoost (Freund and Schapire 97) - 200:31
Basic properties of AdaBoost’s convergence are still not fully understood 01:00
AdaBoost is known for its ability to combine “weak classifiers” into a “strong” classifier - 101:47
AdaBoost is known for its ability to combine “weak classifiers” into a “strong” classifier - 202:46
AdaBoost is known for its ability to combine “weak classifiers” into a “strong” classifier - 303:03
AdaBoost is known for its ability to combine “weak classifiers” into a “strong” classifier - 403:38
Known:04:04
Outline - 105:43
Main Messages06:35
Convergence Rate 107:33
At iteration t ... - 107:43
At iteration t ... - 208:21
At iteration t ... - 308:33
At iteration t ... - 408:47
At iteration t ... - 509:00
At iteration t ... - 609:11
At iteration t ... - 709:18
Theorem 1 - 109:25
Theorem 1 - 210:15
Theorem 1 - 310:53
Theorem 1 - 411:18
Theorem 1 - 511:34
Rate on a Simple Dataset (Log scale)12:01
Outline - 212:31
Theorem 2 - 112:36
Theorem 2 - 213:17
Theorem 2 - 313:37
Theorem 2 - 413:40
Theorem 2 - 513:56
Decomposition Lemma - 114:36
Decomposition Lemma - 215:31
Decomposition Lemma - 315:50
Decomposition Lemma - 416:09
Decomposition Lemma - 516:15
Decomposition Lemma - 616:20
Decomposition Lemma - 716:24
Lemma16:33
To summarize17:29