(Enter summary)
Abstract: A model for on-site learning is presented. The system learns by
querying "hard" patterns while classifying "easy" ones. This model is
related to query-based filtering methods, but takes into account that
in addition to labelling, filtering through the data has a cost.
A few simple policies are introduced and analyzed for a simple
problem (1D high low game). In addition the Query-by-Committee
algorithm (Seung et. al) is suggested as a good approximator of the
model space for real-world domains.... (Update)
Context of citations to this paper: More
.... be selected from the input stream if their information content exceeds a threshold (Seung, Opper, Sompolinsky 1992; Freund et al. 1993; Matan 1995). Alternatively, they may be selected randomly, with some probability proportional to information content, as we do in this...
...the learner knows , it makes sense to use query filtering during the performance task. This results in the following on site model (Matan 1995) that considers the loss associated with classifying or querying and must enforce a policy to minimize the total loss. The system...
Cited by: More
Committee-Based Sample Selection For Probabilistic Classifiers - Argamon-Engelson, Dagan (1999)
(Correct)
Robust Soft-Entropy Neural Network Trees - John
(Correct)
Active Learning During the Performance Task - Matan
(Correct)
Active bibliography (related documents): More All
0.5: Shared Weights Neural Networks in Image Analysis - de Ridder (1996)
(Correct)
0.0: From Conventional Control to Intelligent Neurocontrol.. - RayChaudhuri, Hamey (1995)
(Correct)
0.0: Cost-Effective Querying Leading To Dual Control - RayChaudhuri, Hamey (1996)
(Correct)
Similar documents based on text: More All
0.2: Remote Engineering Tools for the Design of.. - Diachin, Freitag, ..
(Correct)
0.2: State Of The Art (SOTA) Manual For Non-Hazardous Onsite.. - July State Of
(Correct)
0.2: William Forston, Author - Kirk-Mayer Inc Tonopah (1997)
(Correct)
Related documents from co-citation: More All
3: Heterogeneous uncertainty sampling for supervised learning
- Lewis, Catlett - 1994
3: Queries and concept learning (context) - Angluin - 1988
3: Information-based objective functions for active data selection
- MacKay
BibTeX entry: (Update)
Matan, O. 1995. On-site learning. Submitted for publication. http://citeseer.ist.psu.edu/matan95site.html More
@misc{ matan95site,
author = "O. Matan",
title = "site learning",
text = "Matan, O. 1995. On-site learning. Submitted for publication.",
year = "1995",
url = "citeseer.ist.psu.edu/matan95site.html" }
Citations (may not include all citations):
441
Queries and concept learning (context) - Angluin - 1988
317
Learning quickly when irrelevant attributes abound: A new li.. (context) - Littlestone - 1988
217
Artificial Intelligence (context) - Rich, Knight - 1991
131
When networks disagree: Ensemble methods for hybrid neural n..
- Perrone, Cooper - 1993
107
norms : NN pattern classification techniques (context) - Dasarathy, neighbor - 1991
102
Neural network ensembles (context) - Hansen, Salamon - 1990
83
Query by committee
- Seung, Opper et al. - 1992
43
Neural net algorithms that learn in polynomial time from exa.. (context) - Baum - 1991
39
Training connectionist networks with queries and selective s.. (context) - Cohen, Atlas et al. - 1990
29
Measuring the vcdimension of a learning machine
- Vapnik, Levin et al. - 1994
21
Generalization performance of bayes optimal classification a.. (context) - Opper, Haussler - 1991
20
and query by committee (context) - Freund, Seung et al. - 1993
18
Handwritten zip code recognition with multilayer networks (context) - Le Cun, Matan et al. - 1990
2
Handwritten character recognition using neural network archi..
- Matan, Kiang et al. - 1990
Documents on the same site (http://cit.stanford.edu/people/ofer/): More
On Voting Ensembles of Classifiers (Extended Abstract) - Matan
(Correct)
Active Learning During the Performance Task - Matan
(Correct)
Multi-Digit Recognition Using A Space Displacement.. - Matan, Burges, Le Cun.. (1992)
(Correct)
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC