On-line learning algorithms: theory and practice

author: Nicolò Cesa-Bianchi, University of Milano

You might be experiencing some problems with Your Video player.

*Slides*
0:00	On-Line Learning
0:37	Summary
2:54	- Linear classification
2:55	On-line classification
4:22	Linear classifiers
6:46	On-line learning protocol - 1
6:58	On-line learning protocol - 2
7:58	On-line learning protocol - 3
10:18	Remarks
14:17	- The Perceptron algorithm
14:18	Perceptron algorithm - 1
14:50	On-line learning protocol - 2
15:03	Perceptron algorithm - 1
15:24	Perceptron algorithm - 2
17:34	Linear separability - 1
18:19	Linear separability - 2
19:09	Relative loss bound - 1
20:55	Relative loss bound - 2
21:55	Relative loss bound - 3
23:30	Norm of the separator
25:41	Analysis of Perceptron
27:18	Analysis: When a mistake occurs - 1
28:49	Analysis: When a mistake occurs - 2
29:39	When a mistake occurs (cont.) - 1
30:19	When a mistake occurs (cont.) - 2
30:28	When a mistake occurs (cont.) - 3
30:51	The relative mistake bound - 1
31:23	The relative mistake bound - 2
33:17	When a mistake occurs (cont.) - 3
33:39	The relative mistake bound - 2
34:07	- Mistake bounds for separable streams
34:13	Aggressive updates: Hildreth algorithm (1957) - 1
35:01	Aggressive updates: Hildreth algorithm (1957) - 2
35:59	Aggressive updates: Hildreth algorithm (1957) - 3
36:36	Aggressive updates: Hildreth algorithm (1957) - 4
37:09	Aggressive updates: Hildreth algorithm (1957) - 5
37:41	Aggressive updates: Hildreth algorithm (1957) - 6
38:26	Aggressive updates: Hildreth algorithm (1957) - 7
39:07	Analysis for linearly separable streams - 1
39:29	Analysis for linearly separable streams - 2
40:00	Analysis for linearly separable streams - 3
40:16	Analysis (cont.)
40:26	Analysis for linearly separable streams - 3
40:41	Analysis (cont.)
41:20	Aggressive updates: Hildreth algorithm (1957) - 7
41:36	Analysis (cont.)
41:55	Analysis for linearly separable streams - 3
42:03	Analysis (cont.)
45:22	Aggressive updates: Hildreth algorithm (1957) - 7
46:22	Analysis (cont.)
46:34	The cone of consistent hyperplanes - 1
46:41	Analysis (cont.)
47:18	The cone of consistent hyperplanes - 1
48:48	The cone of consistent hyperplanes - 2
48:49	The cone of consistent hyperplanes - 1
48:58	The cone of consistent hyperplanes - 2
49:05	The cone of consistent hyperplanes - 3
49:07	The cone of consistent hyperplanes - 2
49:15	The cone of consistent hyperplanes - 3
50:08	The cone of consistent hyperplanes - 4
50:17	The cone of consistent hyperplanes - 5
50:37	The cone of consistent hyperplanes - 6
51:11	The cone of consistent hyperplanes - 7
51:47	Mistake bounds for various updates - 1
52:03	Mistake bounds for various updates - 2
55:39	Mistake bounds for various updates - 3
55:55	- Online learning and convex optimization
56:05	Aggressive updates for nonseparable streams - 1
57:19	Aggressive updates for nonseparable streams - 2
58:08	Aggressive updates for nonseparable streams - 3
59:05	Aggressive updates for nonseparable streams - 4
59:12	Aggressive updates for nonseparable streams - 5
60:15	SVM and passive-aggressive - 1
61:15	SVM and passive-aggressive - 2
62:06	SVM and passive-aggressive - 3
62:56	SVM and passive-aggressive (cont.) - 1
63:10	SVM and passive-aggressive - 3
63:16	SVM and passive-aggressive (cont.) - 1
63:34	SVM and passive-aggressive (cont.) - 2
64:08	SVM and passive-aggressive (cont.) - 3
64:43	SVM and passive-aggressive (cont.) - 4
64:57	SVM and passive-aggressive (cont.) - 5
65:33	Mistake bounds for PA-I - 1
65:36	SVM and passive-aggressive (cont.) - 5
65:49	SVM and passive-aggressive (cont.) - 2
65:53	Mistake bounds for PA-I - 1
66:25	Mistake bounds for PA-I - 2
66:53	Mistake bounds for PA-I - 3
67:37	Proof of mistake bound for PA-I - 1
67:41	Proof of mistake bound for PA-I - 2
67:53	Proof of mistake bound for PA-I - 3
68:50	Proof of mistake bound for PA-I - 4
71:46	- Kernel-based on-line learning
72:11	On-line learning with kernels - 1
72:25	On-line learning with kernels - 2
72:26	On-line learning with kernels - 1
72:31	On-line learning with kernels - 3
72:34	On-line learning with kernels - 2
72:46	- Kernel-based on-line learning
72:54	On-line learning with kernels - 1
73:17	On-line learning with kernels - 2
73:30	On-line learning with kernels - 3
74:19	On-line learning with kernels - 4
75:32	Kernel Perceptron - 1
75:35	Kernel Perceptron - 2
75:36	Kernel Perceptron - 3
75:54	Kernel Perceptron - 4
75:56	Kernel Perceptron - 5
77:08	Kernel Perceptron - 6
77:40	Memory bounded learning - 1
78:49	Memory bounded learning - 2
79:20	Kernel Perceptron - 6
80:05	Memory bounded learning - 2
81:03	Memory bounded learning - 3
81:21	Memory bounded learning - 4
81:40	A randomized perceptron - 1
81:53	A randomized perceptron - 2
81:58	A randomized perceptron - 3
82:06	A randomized perceptron - 4
82:08	A randomized perceptron - 5
82:10	A randomized perceptron - 6
82:23	A randomized perceptron - 7
82:59	A randomized perceptron - 8
84:16	Empirical performance - stationary
86:49	Empirical performance - nonstationary
87:53	Empirical performance 2nd order - nonstationary

Lecture rating

People found this lecture:
Worth seeing
because it is:
Valuable and informative
Well presented
Easily understandable
Acceptably recorded
You need to login to cast your vote.

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.