Document Update

To update a document of type 'Slides' update the form below and click the buttons Save/Check (if you want to return to this form) or Save/Exit.


Field Name		Field Value
Reference Type		Slides (Audiovisual Material)
Title	(*)(?)
Short Title
Year	(*)
Number of Slides
	Don't forget to press the "+" button to enter the co-author data, if any.		v
1^st Author	(*)(?)
Resume Identifier (1^st Author)	(?)
ORCID (1^st Author)	(?)
Affiliation (1^st Author)	(?)
Author e-Mail Address (1^st Author)
2^nd Author	(*)(?)
Resume Identifier (2^nd Author)	(?)
ORCID (2^nd Author)	(?)
Affiliation (2^nd Author)	(?)
Author e-Mail Address (2^nd Author)
3^rd Author	(*)(?)
Resume Identifier (3^rd Author)	(?)
ORCID (3^rd Author)	(?)
Affiliation (3^rd Author)	(?)
Author e-Mail Address (3^rd Author)
4^th Author	(*)(?)
Resume Identifier (4^th Author)	(?)
ORCID (4^th Author)	(?)
Affiliation (4^th Author)	(?)
Author e-Mail Address (4^th Author)
5^th Author	(*)(?)
Resume Identifier (5^th Author)	(?)
ORCID (5^th Author)	(?)
Affiliation (5^th Author)	(?)
Author e-Mail Address (5^th Author)
Date
Volume
Book Title
Scheduling Information
Content Type	(?)	External Contribution
Keywords	(?)	supervised classifiers image foresting transform image analysis pattern recognition
Abstract	(?)	We present a supervised pattern classifier based on optimum path forest. The samples in a training set are nodes of a complete graph, whose arcs are weighted by the distances between sample feature vectors. The training builds a classifier from key samples (prototypes) of all classes, where each prototype defines an optimum path tree whose nodes are its strongest connected samples. The optimum paths are also considered to label unseen test samples with the classes of their strongest connected prototypes. We show how to find prototypes with none classification errors in the training set and propose a learning algorithm to improve accuracy over an evaluation set. The method is robust to outliers, handles non-separable classes, and can outperform support vector machines.
Subject
Language		Portuguese Spanish French English
Target File
Copyright License	(?)
Rightsholder
Next Higher Units		83LX3pFwXQZ3qyBY/PKn22
Previous Lower Units
File Name	(?)
e-Mail (login)		afalcao@ic.unicamp.br
open \| access \| download \| metadada \| rep: dpi.inpe.br/ismm@80/2007/10.14.17.43
(?)

About Copyrights

If the document that you are submitting is the original (i.e., if it has not yet been published), by submitting it, you are expected to transfer the original document to the owner of the site <mtc-m21b.sid.inpe.br> that will become the copyright holder. The site owner will be responsable for storing, preserving and identifying the original document, authenticating any copies of it, posting it on the web, or transferring it to another site owner with the same duties. You are not supposed to submit the original to another site, but just copies. If the original document definitely ceases to be posted on the web, then it will be considered to be returning to you (i.e., in this case, the copyrights return to the author); for this reason, we recommend that you keep, in your power, a copy of the submitted original.

If the document that you are submitting has already been published (i.e., if it is not the original), this document should be a copy of the original. By submitting it, you are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, you may not submit the document without the explicit permission of the copyright holder.

During the submission period (1), you may update the submitted document or cancel your submission just by substituing the submitted document for a blank one.

(1) Please contact the site administrator <bibdigital@inpe.br> to get information about the submission period.