International Conference on Machine Learning - Bonn 2005

How to Predict Sequences with Bayes, MDL and Experts

author: Marcus Hutter, IDSIA

You might be experiencing some problems with Your Video player.

*Slides*
0:01	Minimum Description Length: Contents
1:55	Minimum Description Length: Abstract
1:57	How to Choose the Prior?
4:36	Indi®erence or Symmetry Principle
7:21	The Maximum Entropy Principle ...
7:23	Occam's Razor \| The Simplicity Principle
7:27	Pre¯x Sets/Codes
9:37	Proof of the Kraft-Inequality
12:10	Priors from Pre¯x Codes
15:18	A Universal Choice of » and M
20:11	The Minimum Description Length Principle
23:54	Predict with Best Model
29:04	Application: Sequence Prediction
31:40	Application: Regression / Polynomial Fitting
33:50	MDL Solution to Polynomial Fitting
37:27	MDL Polynomial Fitting: Determine Degree d
42:00	Minimum Description Length: Summary
44:06	The Similarity Metric: Contents
44:46	The Similarity Metric: Abstract
44:49	Kolmogorov Complexity
47:45	The Universal Similarity Metric
52:32	Tree-Based Clustering
53:24	Genomics & Phylogeny: Mammals
58:20	Genomics & Phylogeny: Mammals
61:36	Genomics & Phylogeny: SARS Virus and Others
61:50	Genomics & Phylogeny: SARS Virus and Others
62:37	Classi¯cation of Di®erent File Types
63:09	Classi¯cation of Di®erent File Types
64:01	Language Tree (Re)construction
64:29	sheme
68:38	Classify Music w.r.t. Composer
69:27	Classify Music w.r.t. Composer
70:19	Further Applications
70:41	The Clustering Method: Summary
72:16	Prediction with Expert Advice: Contents
86:43	Prediction with Expert Advice: Abstract
86:45	Prediction with Expert Advice (PEA) - Informal
87:34	Prediction with Expert Advice (PEA) - Setup
87:49	Goals
89:34	Best Expert in Hindsight (BEH)
89:46	Weighted Majority (WM)
90:14	Follow the Perturbed Leader (FPL)
90:54	Regret Bounds for n < 1 and ki = ln n
92:12	Regret Bounds for n = 1 and general ki
92:21	Some more FPL Results
92:23	PEA versus Bayes Bounds { Formal
93:23	Naive Ansatz: Follow the Leader (FL)

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.