Incremental and Decremental Training for Linear Classification thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Incremental and Decremental Training for Linear Classification

Published on Oct 07, 20141792 Views

In classification, if a small number of instances is added or removed, incremental and decremental techniques can be applied to quickly update the model. However, the design of incremental and decreme

Related categories

Chapter list

Incremental and Decremental Training for Linear Classi cation00:00
Outline - 100:10
Outline - 200:15
Incremental and Decremental Training00:16
Why Linear Classi cation for Incremental and Decremental learning?00:58
Linear Classi cation01:46
Dual Problems02:26
Incremental and Decremental Learning with Warm Start (1/2)02:39
Incremental and Decremental Learning with Warm Start (2/2)03:09
Outline - 304:17
Analysis on Our Setting04:24
Outline - 404:41
Primal or Dual Problem: Which is Better? 04:43
Primal Initial Objective Value for Incremental Learning - 105:11
Primal Initial Objective Value for Incremental Learning - 205:19
Primal Initial Objective Value for Incremental Learning - 305:23
Dual Initial Objective Value for Incremental Learning - 105:31
Dual Initial Objective Value for Incremental Learning - 205:47
Comparison of Primal and Dual Initial Objective Values (1/2)06:03
Comparison of Primal and Dual Initial Objective Values (2/2)07:03
Decremental Learning Initial Objective Values07:21
Outline - 507:55
Optimization Methods with Warm Start07:58
Optimization Methods in Experiments09:01
Outline - 609:44
Additional Information Requirement09:47
Outline - 710:42
Experiment Setting10:43
Data Sets11:17
Relative Di erence to Optimal Objective Value for Incremental Learning11:36
Incremental Learning12:25
Decremental Learning13:23
Outline - 813:28
Conclusions13:31