2016
|
Monteiro, C; Mathew, C; Shipman, F; Gutierrez-Osuna, R Detecting and Identifying Sign Languages through Visual Features Inproceedings In: 2016 IEEE International Symposium on Multimedia (ISM), 2016. @inproceedings{monteiro2016ism,
title = {Detecting and Identifying Sign Languages through Visual Features},
author = {C Monteiro and C Mathew and F Shipman and R Gutierrez-Osuna},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/monteiro2016ism.pdf},
year = {2016},
date = {2016-12-11},
booktitle = {2016 IEEE International Symposium on Multimedia (ISM)},
keywords = {Computer vision, Gestures, Pattern recognition},
pubstate = {published},
tppubtype = {inproceedings}
}
|
2015
|
Shipman, F; Gutierrez-Osuna, R; Shipman, T; Monteiro, C; Karappa, V Towards a Distributed Digital Library for Sign Language Content Inproceedings In: Proc. ACM/IEEE Joint Conference on Digital Libraries, 2015. @inproceedings{shipman2015jcdl,
title = {Towards a Distributed Digital Library for Sign Language Content},
author = {F Shipman and R Gutierrez-Osuna and T Shipman and C Monteiro and V Karappa},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/shipman2015jcdl.pdf},
year = {2015},
date = {2015-04-02},
booktitle = {Proc. ACM/IEEE Joint Conference on Digital Libraries},
volume = {in press},
keywords = {Computer vision, Gestures, Pattern recognition},
pubstate = {published},
tppubtype = {inproceedings}
}
|
Gupta, A; Gutierrez-Osuna, R; Christy, M; Capitanu, B; Auvil, L; Grumbach, L; Furuta, R; Mandell, L Automatic assessment of OCR quality in historical documents Inproceedings In: Proc. AAAI, 2015. @inproceedings{gupta2015aaai,
title = {Automatic assessment of OCR quality in historical documents},
author = {A Gupta and R Gutierrez-Osuna and M Christy and B Capitanu and L Auvil and L Grumbach and R Furuta and L Mandell},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/gupta2015aaai.pdf},
year = {2015},
date = {2015-01-25},
booktitle = {Proc. AAAI},
volume = {in press},
keywords = {OCR, Other, Pattern recognition},
pubstate = {published},
tppubtype = {inproceedings}
}
|
2014
|
Christy, M; Auvil, L; Gutierrez-Osuna, R; Capitanu, B; Gupta, A; Grumbach, E Diagnosing Page Image Problems with Post-OCR Triage for eMOP Inproceedings In: Proc. Digital Humanities Conference, 2014. @inproceedings{christy2014emopdhc,
title = {Diagnosing Page Image Problems with Post-OCR Triage for eMOP},
author = {M Christy and L Auvil and R Gutierrez-Osuna and B Capitanu and A Gupta and E Grumbach},
year = {2014},
date = {2014-07-08},
booktitle = {Proc. Digital Humanities Conference},
volume = {in press},
keywords = {OCR, Other, Pattern recognition},
pubstate = {published},
tppubtype = {inproceedings}
}
|
Perera-Lluna, A; Manivannan, K; Xu, P; Gutierrez-Osuna, R; Benner, C; Russell, B D Automatic Capacitor Bank Identification in Power Distribution Systems Journal Article In: Electric Power Systems Research, vol. in press, 2014. @article{alex2014epsr,
title = {Automatic Capacitor Bank Identification in Power Distribution Systems},
author = {A Perera-Lluna and K Manivannan and P Xu and R Gutierrez-Osuna and C Benner and B D Russell},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/alex2014epsr.pdf},
year = {2014},
date = {2014-02-06},
journal = {Electric Power Systems Research},
volume = {in press},
keywords = {Other, Pattern recognition},
pubstate = {published},
tppubtype = {article}
}
|
Yu, Y; Choe, Y; Gutierrez-Osuna, R Context-sensitive Intra-class Clustering Journal Article In: Pattern Recognition Letters, vol. 37, pp. 85-93, 2014. @article{yingwei2013prl,
title = {Context-sensitive Intra-class Clustering},
author = {Y Yu and Y Choe and R Gutierrez-Osuna},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/yingwei2013prl.pdf},
year = {2014},
date = {2014-02-01},
journal = {Pattern Recognition Letters},
volume = {37},
pages = {85-93},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {article}
}
|
2007
|
Koh, E; Caruso, D; Kerne, A; Gutierrez-Osuna, R Elimination of junk document surrogate candidates through pattern recognition Conference Proceedings of the 2007 ACM symposium on Document engineering, ACM 2007. @conference{koh2007elimination,
title = {Elimination of junk document surrogate candidates through pattern recognition},
author = {E Koh and D Caruso and A Kerne and R Gutierrez-Osuna},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/koh2007elimination.pdf},
year = {2007},
date = {2007-01-01},
booktitle = {Proceedings of the 2007 ACM symposium on Document engineering},
pages = {187--195},
organization = {ACM},
abstract = {A surrogate is an object that stands for a document and enables navigation to that document. Hypermedia is often represented with textual surrogates, even though studies have shown that image and text surrogates facilitate the formation of mental models and overall understanding. Surrogates may be formed by breaking a document down into a set of smaller elements, each of which is a surrogate candidate. While processing these surrogate candidates from an HTML document, relevant information may appear together with less useful junk material, such as navigation bars and advertisements.
This paper develops a pattern recognition based approach for eliminating junk while building the set of surrogate candidates. The approach defines features on candidate elements, and uses classification algorithms to make selection decisions based on these features. For the purpose of defining features in surrogate candidates, we introduce the Document Surrogate Model (DSM), a streamlined Document Object Model (DOM)-like representation of semantic structure. Using a quadratic classifier, we were able to eliminate junk surrogate candidates with an average classification rate of 80%. By using this technique, semiautonomous agents can be developed to more effectively generate surrogate collections for users. We end by describing a new approach for hypermedia and the semantic web, which uses the DSM to define value-added surrogates for a document.},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {conference}
}
A surrogate is an object that stands for a document and enables navigation to that document. Hypermedia is often represented with textual surrogates, even though studies have shown that image and text surrogates facilitate the formation of mental models and overall understanding. Surrogates may be formed by breaking a document down into a set of smaller elements, each of which is a surrogate candidate. While processing these surrogate candidates from an HTML document, relevant information may appear together with less useful junk material, such as navigation bars and advertisements.
This paper develops a pattern recognition based approach for eliminating junk while building the set of surrogate candidates. The approach defines features on candidate elements, and uses classification algorithms to make selection decisions based on these features. For the purpose of defining features in surrogate candidates, we introduce the Document Surrogate Model (DSM), a streamlined Document Object Model (DOM)-like representation of semantic structure. Using a quadratic classifier, we were able to eliminate junk surrogate candidates with an average classification rate of 80%. By using this technique, semiautonomous agents can be developed to more effectively generate surrogate collections for users. We end by describing a new approach for hypermedia and the semantic web, which uses the DSM to define value-added surrogates for a document. |
2005
|
Yu, Y; Gutierrez-Osuna, R; Choe, Y Preserving Class Discriminatory Information by Context-Sensitve Intra-Class Clustering Algorithm Technical Report 2005. @techreport{yu2005techreport,
title = {Preserving Class Discriminatory Information by Context-Sensitve Intra-Class Clustering Algorithm},
author = {Y Yu and R Gutierrez-Osuna and Y Choe},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/yu05techreport.pdf},
year = {2005},
date = {2005-02-25},
abstract = {Many powerful techniques in supervised learning (e.g. linear discriminant analysis, LDA, and quadratic classifier) assume that data in each class have a single Gaussian distribution. In reality, data in the class of interest, i.e., the object class, could have non-Gaussian distributions and could be isolated into several subgroups by the data from other classes (the context classes). To address this problem, one possible way is to partition one class into several subclasses. This intra-class clustering should depend on the data structure of the class of interest (object class) as well as the distributions of all other classes (context classes). In this paper, we presented a novel method of intra-class clustering which can divide a non-Gaussian class data into several Gaussian-like clusters, and at the same time this algorithm is context sensitive, which can maximally reduce the overlapping among resulting classes and also between the object class and the context classes. The method can serve as a general data preprocessing method to improve performance of supervised learning algorithms such as LDA and quadratic classifiers.},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {techreport}
}
Many powerful techniques in supervised learning (e.g. linear discriminant analysis, LDA, and quadratic classifier) assume that data in each class have a single Gaussian distribution. In reality, data in the class of interest, i.e., the object class, could have non-Gaussian distributions and could be isolated into several subgroups by the data from other classes (the context classes). To address this problem, one possible way is to partition one class into several subclasses. This intra-class clustering should depend on the data structure of the class of interest (object class) as well as the distributions of all other classes (context classes). In this paper, we presented a novel method of intra-class clustering which can divide a non-Gaussian class data into several Gaussian-like clusters, and at the same time this algorithm is context sensitive, which can maximally reduce the overlapping among resulting classes and also between the object class and the context classes. The method can serve as a general data preprocessing method to improve performance of supervised learning algorithms such as LDA and quadratic classifiers. |
2003
|
Florkey, L; Gutierrez-Osuna, R; Bullock, J D Computer-Aided Pattern Recognition in the Classification of Superficial Corneal Injuries Conference Proceedings of the Annual Meeting of the American Academy of Optometry, 2003. @conference{florkey03corneal,
title = {Computer-Aided Pattern Recognition in the Classification of Superficial Corneal Injuries},
author = {L Florkey and R Gutierrez-Osuna and J D Bullock},
year = {2003},
date = {2003-12-04},
booktitle = {Proceedings of the Annual Meeting of the American Academy of Optometry},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {conference}
}
|
Mousavi, M J; Butler-Purry, K L; Gutierrez-Osuna, R; Najafi, M Classification of load change transients and incipient abnormalities in underground cable using pattern analysis techniques Conference Proceedings of 2003 IEEE PES Transmission and Distribution Conference and Exposition, IEEE 2003. @conference{mousavi2003classification,
title = {Classification of load change transients and incipient abnormalities in underground cable using pattern analysis techniques},
author = {M J Mousavi and K L Butler-Purry and R Gutierrez-Osuna and M Najafi},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/mousavi2003tdce.pdf},
year = {2003},
date = {2003-01-01},
booktitle = {Proceedings of 2003 IEEE PES Transmission and Distribution Conference and Exposition},
pages = {175--180},
organization = {IEEE},
abstract = {This paper presents a feasibility study on the application of pattern analysis techniques to classify load change transients and incipient abnormalities in an underground distribution cable lateral. The data were collected using an on-line monitoring system installed in a residential area in Dallas. A set of features obtained from wavelet packet analysis was evaluated. Methods of dimensionality reduction were employed to overcome the curse of dimensionality while preserving a good classification rate. The classification results using k-nearest-neighbor (KNN) classifiers show that the proposed methodology can be used to classify load change transients and incipient abnormalities.},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {conference}
}
This paper presents a feasibility study on the application of pattern analysis techniques to classify load change transients and incipient abnormalities in an underground distribution cable lateral. The data were collected using an on-line monitoring system installed in a residential area in Dallas. A set of features obtained from wavelet packet analysis was evaluated. Methods of dimensionality reduction were employed to overcome the curse of dimensionality while preserving a good classification rate. The classification results using k-nearest-neighbor (KNN) classifiers show that the proposed methodology can be used to classify load change transients and incipient abnormalities. |
Bryll, R; Gutierrez-Osuna, R; Quek, F Attribute bagging: improving accuracy of classifier ensembles by using random feature subsets Journal Article In: Pattern Recognition, vol. 36, no. 6, pp. 1291–1302, 2003. @article{bryll2003attribute,
title = {Attribute bagging: improving accuracy of classifier ensembles by using random feature subsets},
author = {R Bryll and R Gutierrez-Osuna and F Quek},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/bryll2003attribute-1.pdf},
year = {2003},
date = {2003-01-01},
journal = {Pattern Recognition},
volume = {36},
number = {6},
pages = {1291--1302},
publisher = {Elsevier},
abstract = {We present attribute bagging (AB), a technique for improving the accuracy and stability ofclassifierensembles induced using randomsubsets of features. AB is a wrapper method that can be used with any learning algorithm. It establishes an appropriate attributesubset size and then randomly selectssubsets of features, creating projections of the training set on which the ensembleclassifiers are built. The induced classifiers are then used for voting. This article compares the performance of our AB method with bagging and other algorithms on a hand-pose recognition dataset. It is shown that AB gives consistently better results than bagging, both in accuracy and stability. The performance of ensemble voting in bagging and the AB method as a function of the attributesubset size and the number of voters for both weighted and unweighted voting is tested and discussed. We also demonstrate that ranking the attributesubsets by their classification accuracy and voting using only the best subsets further improves the resulting performance of the ensemble.},
keywords = {Pattern recognition},
pubstate = {published},
tppubtype = {article}
}
We present attribute bagging (AB), a technique for improving the accuracy and stability ofclassifierensembles induced using randomsubsets of features. AB is a wrapper method that can be used with any learning algorithm. It establishes an appropriate attributesubset size and then randomly selectssubsets of features, creating projections of the training set on which the ensembleclassifiers are built. The induced classifiers are then used for voting. This article compares the performance of our AB method with bagging and other algorithms on a hand-pose recognition dataset. It is shown that AB gives consistently better results than bagging, both in accuracy and stability. The performance of ensemble voting in bagging and the AB method as a function of the attributesubset size and the number of voters for both weighted and unweighted voting is tested and discussed. We also demonstrate that ranking the attributesubsets by their classification accuracy and voting using only the best subsets further improves the resulting performance of the ensemble. |
Courte, D E; Rizki, M M; Tamburino, L A; Gutierrez-Osuna, R Evolutionary optimization of Gaussian windowing functions for data preprocessing Journal Article In: International Journal on Artificial Intelligence Tools, vol. 12, no. 1, pp. 17–36, 2003. @article{courte2003evolutionary,
title = {Evolutionary optimization of Gaussian windowing functions for data preprocessing},
author = {D E Courte and M M Rizki and L A Tamburino and R Gutierrez-Osuna},
url = {https://psi.engr.tamu.edu/wp-content/uploads/2018/01/courte2003evolutionary-1.pdf},
year = {2003},
date = {2003-01-01},
journal = {International Journal on Artificial Intelligence Tools},
volume = {12},
number = {1},
pages = {17--36},
publisher = {WORLD SCIENTIFIC PUBLISHING},
abstract = {The average classification accuracy of an odor classification system is improved using a genetic algorithm to determine optimal parameters for feature extraction. Gaussian windowing functions, called "kernels" are evolved to extract information from the transient response of an array of gas sensors, resulting in a reduced set of extracted features for a linear discriminant pattern classification system. Results show significant improvements are achieved when compared to results obtained using a predetermined and fixed set of four bell-shaped kernels for every sensor. Examination of the evolved kernels reveals the areas of the sensor responses where discriminating information was identified. A novel data migration approach during training helps prevent overtraining, and the fitness measure chosen incorporates adjustments for both population diversity and solution complexity. A variety of adjustable parameters, including the definition of a time-varying dynamic weighting factor, encourage experimentation in order to appropriately tune the sampling methods and fitness measure.},
keywords = {Electronic nose, Pattern recognition},
pubstate = {published},
tppubtype = {article}
}
The average classification accuracy of an odor classification system is improved using a genetic algorithm to determine optimal parameters for feature extraction. Gaussian windowing functions, called "kernels" are evolved to extract information from the transient response of an array of gas sensors, resulting in a reduced set of extracted features for a linear discriminant pattern classification system. Results show significant improvements are achieved when compared to results obtained using a predetermined and fixed set of four bell-shaped kernels for every sensor. Examination of the evolved kernels reveals the areas of the sensor responses where discriminating information was identified. A novel data migration approach during training helps prevent overtraining, and the fitness measure chosen incorporates adjustments for both population diversity and solution complexity. A variety of adjustable parameters, including the definition of a time-varying dynamic weighting factor, encourage experimentation in order to appropriately tune the sampling methods and fitness measure. |