Results 1 -
8 of
8
Indexing of Compressed Time Series
- Data Mining in Time Series Databases
"... We describe a procedure for identifying major minima and maxima of a time series, and present two applications of this procedure. The first application is fast compression of a series, by selecting major extrema and discarding the other points. The compression algorithm runs in linear time and takes ..."
Abstract
-
Cited by 13 (3 self)
- Add to MetaCart
We describe a procedure for identifying major minima and maxima of a time series, and present two applications of this procedure. The first application is fast compression of a series, by selecting major extrema and discarding the other points. The compression algorithm runs in linear time and takes constant memory. The second application is indexing of compressed series by their major extrema, and retrieval of series similar to a given pattern. The retrieval procedure searches for the series whose compressed representation is similar to the compressed pattern. It allows the user to control the trade-off between the speed and accuracy of retrieval. We show the effectiveness of the compression and retrieval for stock charts, meteorological data, and electroencephalograms. Keywords. Time series, compression, fast retrieval, similarity measures. 1
End-user feature labeling: A locally-weighted regression approach
- In Proc. IUI, ACM
, 2011
"... When intelligent interfaces, such as intelligent desktop assistants, email classifiers, and recommender systems, customize themselves to a particular end user, such customizations can decrease productivity and increase frustration due to inaccurate predictions—especially in early stages, when traini ..."
Abstract
-
Cited by 8 (6 self)
- Add to MetaCart
(Show Context)
When intelligent interfaces, such as intelligent desktop assistants, email classifiers, and recommender systems, customize themselves to a particular end user, such customizations can decrease productivity and increase frustration due to inaccurate predictions—especially in early stages, when training data is limited. The end user can improve the learning algorithm by tediously labeling a substantial amount of additional training data, but this takes time and is too ad hoc to target a particular area of inaccuracy. To solve this problem, we propose a new learning algorithm based on locally weighted regression for feature labeling by end users, enabling them to point out which features are important for a class, rather than provide new training instances. In our user study, the first allowing ordinary end users to freely choose features to label directly from text documents, our algorithm was both more effective than others at leveraging end users ’ feature labels to improve the learning algorithm, and more robust to real users ’ noisy feature labels. These results strongly suggest that allowing users to freely choose features to label is a promising method for allowing end users to improve learning algorithms effectively. Author Keywords Feature labeling, locally weighted logistic regression, machine
ENSEMBLE OF FEATURE SELECTION TECHNIQUES FOR HIGH DIMENSIONAL DATA
, 2012
"... ACKNOWLEDGEMENTS This thesis would not have been possible without the guidance and the help of several individuals who in one way or another contributed and extended their valuable assistance in the preparation and completion of this study. I especially want to thank my professor, Dr. Huanjing Wang ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
(Show Context)
ACKNOWLEDGEMENTS This thesis would not have been possible without the guidance and the help of several individuals who in one way or another contributed and extended their valuable assistance in the preparation and completion of this study. I especially want to thank my professor, Dr. Huanjing Wang for her guidance during my thesis. Her perpetual energy and enthusiasm motivated me to put in my best effort. In addition, she was always accessible and willing to help me with all the queries and difficulties I faced. It would not have been possible to write this thesis if Dr. Guangming Xing hadn’t kindled an interest in me to start my thesis. My utmost gratitude goes to him whose sincerity and encouragement has been my driving force. My thesis found a strong foundation and gained strength from the lectures and seminars given by Dr. Qi Li. He gave in valuable inputs for my thesis and guided me. His
End-User Feature Labeling via Locally Weighted Logistic Regression
- in: Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence, Special Track on New Scientific and Technical Advances in ResearchNew Scientific and Technical Advances in Research, AAAI
, 2011
"... Applications that adapt to a particular end user often make inaccurate predictions during the early stages when training data is limited. Although an end user can improve the learning algorithm by labeling more training data, this process is time consuming and too ad hoc to target a particular area ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
Applications that adapt to a particular end user often make inaccurate predictions during the early stages when training data is limited. Although an end user can improve the learning algorithm by labeling more training data, this process is time consuming and too ad hoc to target a particular area of inaccuracy. To solve this problem, we propose a new learning algorithm based on Locally Weighted Logistic Regression for feature labeling by end users, enabling them to point out which features are important for a class, rather than provide new training instances. In our user study, the first allowing ordinary end users to freely choose features to label directly from text documents, our algorithm was more effective than others at leveraging end users ’ feature labels to improve the learning algorithm. Our results strongly suggest that allowing users to freely choose features to label is a promising method for allowing end users to improve learning algorithms effectively.
Content-Based Image Classification A Non-Parametric Classification
"... The conclusion of this work represents the end of a cycle that was only made possible with the support of various people, to whom I express my gratitude. First of all, I would like to thank Prof. Mário Figueiredo and Prof. Pedro Aguiar for having accepted to supervise my work and for their cooperati ..."
Abstract
- Add to MetaCart
(Show Context)
The conclusion of this work represents the end of a cycle that was only made possible with the support of various people, to whom I express my gratitude. First of all, I would like to thank Prof. Mário Figueiredo and Prof. Pedro Aguiar for having accepted to supervise my work and for their cooperation. I also thank them for having given me time to develop and define my research direction. I have to thank Prof. José António Santos for the great patience and effort to rectify my English writings. On top of all, I want to thank my family, especially my parents, Virgílio Ferreira and Maria Anselmo Ferreira, who did everything to give me the wholehearted support and the conditions necessary to my personal and academic growth. I also thank my brother, Samuel M. Ferreira, for his encouragement, organization and methodology advice, that moved me. I show gratitude to my sister-in-law, Joana M. Ferreira, and my beloved niece and nephew, Maria M. Ferreira and Mateus M. Ferreira, for the joy they infected me with. Obviously, I am grateful to the rest of my extended family for their unconditional support in difficult moments. I thank all the lads with special highlights to my good friends Luís Oliveira, João Nuno Santos, Ivo Anastácio and Pedro Martins for their help and contribution to surpass all the difficulties associated with a master’s course. Finally, I would like to thank all the friends, teachers and institutions (IST and ESL) that throughout my life worried about my scholar success.
1 Content-Based Image Classification: A Non-Parametric Approach
"... Abstract — The rise of the amount imagery on the Internet, as well as in multimedia systems, has motivated research work on visual information retrieval (VIR) systems and on automatic analysis of image databases. In this work, we develop a classification system that allows to recognize and recover t ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract — The rise of the amount imagery on the Internet, as well as in multimedia systems, has motivated research work on visual information retrieval (VIR) systems and on automatic analysis of image databases. In this work, we develop a classification system that allows to recognize and recover the class of a query image based on its content. Such systems are called Content-Based Image Retrieval (CBIR). CBIR systems describe each image (either the query or the ones in the database) by a set of features that are automatically extracted. Then, the feature vectors are fed into a classifier. In this thesis, the processes of image feature selection and extraction uses descriptors and techniques such as Scale
Supervised and Semi-supervised Approaches Based on Locally-Weighted Logistic Regression 1
"... Permanent City Research Online ..."
(Show Context)
