Add to Bookmark
|
Make VideoWired.com My Homepage
|
VideoWired Tools
VIDEO SEARCH
HOME
MUSIC
SPORTS
COMEDY
STUNTS
ENTERTAINMENT
NEWS
TV
NEW VIDEOS
CHANNELS
FAVORITE VIDEOS
MOST VIEWED
Categories
Home
Music
Sports
Comedy
Stunts
Entertainment
News
Television
Channels
StupidVideos.com
MySpace
Heavy
YouTube
WheelsTV
Maxim Magazine
Access Hollywood
E! Online
AOL News
AOL Music
MSN Video
TMZ.com
Oxygen
Oscar.com
Showtime
Fuel TV
EVTV1.com
Speed
NBA.com
NFL.com
NBC Sports
Fox Sports
RooTV
CBS SportsLine
CBS
CNN
NBC
Yahoo News
Business Week
Reuters
Weather Channel
Issues in Evaluation of Stream Learning Algorithms Details
Issues in Evaluation of Stream Learning Algorithms
Learning from data streams is a research area of increasing importance. Nowadays, several stream learning algorithms have been developed. Most of them learn decision models that continuously evolve over time, run in resource-aware environments, detect and react to changes in the environment generating data. One important issue, not yet conveniently addressed, is the design of experimental work to evaluate and compare decision models that evolve over time. There are no golden standards for assessing performance in non-stationary environments. This paper proposes a general framework for assessing predictive stream learning algorithms. We defend the use of Predictive Sequential methods for error estimate — the prequential error. The prequential error allows us to monitor the evolution of the performance of models that evolve over time. Nevertheless, it is known to be a pessimistic estimator in comparison to holdout estimates. To obtain more reliable estimators we need some forgetting mechanism. Two viable alternatives are: sliding windows and fading factors. We observe that the prequential error converges to an holdout estimator when estimated over a sliding window or using fading factors. %A similar observation applies for fading factors. We present illustrative examples of the use of prequential error estimators, using fading factors, for the tasks of: i. assessing performance of a learning algorithm; ii. comparing learning algorithms; iii. hypothesis testing using McNemar test; and iv. change detection using Page-Hinkley test. In these tasks, the prequential error estimated using fading factors provide reliable estimators. In comparison to sliding windows, fading factors are faster and memory-less, a requirement for streaming applications. This paper is a contribution to a discussion in the good-practices on performance assessment when learning dynamic models that evolve over time.
Channel:
VideoLectures
Video Length:
0s
Date Found:
Wed, Oct 13 2010 5:18 PM
Category:
Educational
Date Produced:
View Count:
0
Advertisement
Advertisement
Related Results From Educational
Healing and Blessing wrapped up in Gods gift of Righteousness.avi
From:
veehd
on
Sun, Jul 17 2011 10:53 PM
Sugar Highs and Lows: The New Science of Sugar Addiction
From:
UCTV
on
Sat, Jul 16 2011 5:10 PM
Roving Mars: Spirit, Opportunity, and the Exploration of the Red Planet
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Research on Aging: The Opportunity for Health Care via Hospice and Palliative Care
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Conversations with History: The Future of Economic Growth in a Multispeed World with Michael Spence
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
CARTA: The Genetics of Humanness: The Neanderthal and Denisovan Genomes / The Genetics of Human and Ape Stem Cells
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Children and Armed Conflict: The International Response
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Bruin Talk - July 2011 (1)
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Approaches to Care Giving
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Basic Immunology: Nuts and Bolts of the Immune System
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
The Future of Planetary Exploration
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Sugar Highs and Lows: How to Have a Sweet Ending - Interventions to Reduce Sugar Consumption
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
Osher UCSD: Redrawing Lines Between Chimps and Humans
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
UCSD at 50: A Legacy of the Future
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
CARTA: The Genetics of Humanness: Gene Duplications and Deletions and Accelerated Regions in the Human Genome
From:
UCTV
on
Sat, Jul 16 2011 5:09 PM
EurekaFest 2011 - Beijing No. 4 High School InvenTeam
From:
MIT TechTV
on
Sat, Jul 16 2011 3:22 PM
A student from the Beijing No. 4 High School InvenTeam explains their team’s invention, a classroom desk that protects students in an earthquake.
2011 Caribbean Little League Championships
From:
Livestream
on
Sat, Jul 16 2011 1:26 PM
Channel 12 is the PBS affiliate that serves the US Virgin Islands. WTJX operates on four channels, each with distinct programming: Original, Kids, Signature, & Education.
2011 Caribbean Little League Championships
From:
Livestream
on
Sat, Jul 16 2011 1:26 PM
Channel 12 is the PBS affiliate that serves the US Virgin Islands. WTJX operates on four channels, each with distinct programming: Original, Kids, Signature, & Education.
Getting Started with Fireworks: Intro to Using the Slicing Tool
From:
iTunes
on
Sat, Jul 16 2011 12:20 PM
In this video tutorial, Connor Turnbull will show us around the basics for using the Slicing tool in Adobe Fireworks. This is part of a series of tutorials that look at the unique features in Fireworks that make it such a great tool for web designers.
Fireworks Pro Series - Batch Processing
From:
iTunes
on
Sat, Jul 16 2011 12:20 PM
Web designers talk a lot about features or gadgets that are "magic wands"... but today we’re going to look at a unique feature inside Adobe Fireworks that might even live up to the name. If you haven't taken the time to check out the "batch processing" feature in Adobe Fireworks yet, no...
VIDEOWIRED.COM FEATURED
Watch this week’s W...
Channel:
CBS News
Apartment buildings...
Channel:
CBS News
Murdochs testify, ...
Channel:
CBS News
Retired NFL players...
Channel:
CBS News
Laser brain surgery...
Channel:
KTRK Houston
Can Murdoch maintai...
Channel:
CBS News
Gov. Scott Walker g...
Channel:
MSNBC
News Corp. no diffe...
Channel:
MSNBC
Bachmann: Migraines...
Channel:
CBS News
AZ Rep. Trent Frank...
Channel:
MSNBC
Preview: The One Wh...
Channel:
CBS News
Jim Lee: Comics "ve...
Channel:
CBS News