Publications
| 1. | Wilkinson, Bryan; Oates, Tim: A Gold Standard for Scalar Adjectives. Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016), European Language Resources Association (ELRA), Portorož, Slovenia, 2016. (Type: Conference | Links | BibTeX) @conference{Wilkinson2016, title = {A Gold Standard for Scalar Adjectives}, author = {Bryan Wilkinson and Tim Oates }, url = {http://www.lrec-conf.org/proceedings/lrec2016/pdf/1004_Paper.pdf}, year = {2016}, date = {2016-05-23}, booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016)}, publisher = {European Language Resources Association (ELRA)}, address = {Portorož, Slovenia}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
| 2. | Zhiguang Wang Tim Oates, James Lo : Adaptive Normalized Risk-Averting Training for Deep Neural Networks. In: Proceedings of The Thirtieth AAAI Conference on Artificial Intelligence, AAAI, 2016. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wang2016adaptive, title = {Adaptive Normalized Risk-Averting Training for Deep Neural Networks}, author = {Zhiguang Wang, Tim Oates, James Lo}, url = {http://www.aaai.org/ocs/index.php/AAAI/AAAI16/paper/view/11765}, year = {2016}, date = {2016-02-17}, booktitle = {Proceedings of The Thirtieth AAAI Conference on Artificial Intelligence}, publisher = {AAAI}, abstract = {This paper proposes a set of new error criteria and a learning approach, called Adaptive Normalized Risk-Averting Training (ANRAT) to attack the non-convex optimization problem in training deep neural networks without pretraining. Theoretically, we demonstrate its effectiveness based on the expansion of the convexity region. By analyzing the gradient on the convexity index $\lambda$, we explain the reason why our learning method using gradient descent works. In practice, we show how this training method is successfully applied for improved training of deep neural networks to solve visual recognition tasks on the MNIST and CIFAR-10 datasets. Using simple experimental settings without pretraining and other tricks, we obtain results comparable or superior to those reported in recent literature on the same tasks using standard ConvNets + MSE/cross entropy. Performance on deep/shallow multilayer perceptron and Denoised Auto-encoder is also explored. ANRAT can be combined with other quasi-Newton training methods, innovative network variants, regularization techniques and other common tricks in DNNs. Other than unsupervised pretraining, it provides a new perspective to address the non-convex optimization strategy in training DNNs.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper proposes a set of new error criteria and a learning approach, called Adaptive Normalized Risk-Averting Training (ANRAT) to attack the non-convex optimization problem in training deep neural networks without pretraining. Theoretically, we demonstrate its effectiveness based on the expansion of the convexity region. By analyzing the gradient on the convexity index $lambda$, we explain the reason why our learning method using gradient descent works. In practice, we show how this training method is successfully applied for improved training of deep neural networks to solve visual recognition tasks on the MNIST and CIFAR-10 datasets. Using simple experimental settings without pretraining and other tricks, we obtain results comparable or superior to those reported in recent literature on the same tasks using standard ConvNets + MSE/cross entropy. Performance on deep/shallow multilayer perceptron and Denoised Auto-encoder is also explored. ANRAT can be combined with other quasi-Newton training methods, innovative network variants, regularization techniques and other common tricks in DNNs. Other than unsupervised pretraining, it provides a new perspective to address the non-convex optimization strategy in training DNNs. |
| 3. | Clemens, John: Automatic Classification of Object Code using Machine Learning. In: Digital Investigation , pp. S156 - S162, 2015, ISSN: 1742-2876, (DFRWS USA 2015). (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{Clemens2015, title = {Automatic Classification of Object Code using Machine Learning}, author = {John Clemens}, url = {http://www.sciencedirect.com/science/article/pii/S1742287615000523 http://www.sciencedirect.com/science/article/pii/S1742287615000523/pdfft?md5=a270d4d9b1816b67a51e59ba3e6e2284&pid=1-s2.0-S1742287615000523-main.pdf}, doi = {10.1016/j.diin.2015.05.007}, issn = {1742-2876}, year = {2015}, date = {2015-08-09}, urldate = {2015-10-08}, booktitle = {Digital Investigation }, issuetitle = {The Proceedings of the Fifteenth Annual DFRWS Conference }, journal = {Digital Investigation }, volume = {14, Supplement 1}, pages = {S156 - S162}, abstract = {Abstract Recent research has repeatedly shown that machine learning techniques can be applied to either whole files or file fragments to classify them for analysis. We build upon these techniques to show that for samples of un-labeled compiled computer object code, one can apply the same type of analysis to classify important aspects of the code, such as its target architecture and endianess. We show that using simple byte-value histograms we retain enough information about the opcodes within a sample to classify the target architecture with high accuracy, and then discuss heuristic-based features that exploit information within the operands to determine endianess. We introduce a dataset with over 16000 code samples from 20 architectures and experimentally show that by using our features, classifiers can achieve very high accuracy with relatively small sample sizes. }, note = {DFRWS USA 2015}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Abstract Recent research has repeatedly shown that machine learning techniques can be applied to either whole files or file fragments to classify them for analysis. We build upon these techniques to show that for samples of un-labeled compiled computer object code, one can apply the same type of analysis to classify important aspects of the code, such as its target architecture and endianess. We show that using simple byte-value histograms we retain enough information about the opcodes within a sample to classify the target architecture with high accuracy, and then discuss heuristic-based features that exploit information within the operands to determine endianess. We introduce a dataset with over 16000 code samples from 20 architectures and experimentally show that by using our features, classifiers can achieve very high accuracy with relatively small sample sizes. |
| 4. | Zhiguang wang, Tim Oates : Imaging Time-Series to Improve Classification and Imputation. In: Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI), pp. 3939-3945, AAAI, 2015. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wang2015imaging, title = {Imaging Time-Series to Improve Classification and Imputation}, author = {Zhiguang wang, Tim Oates}, url = {http://ijcai.org/papers15/Papers/IJCAI15-553.pdf}, year = {2015}, date = {2015-08-01}, booktitle = {Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI)}, pages = {3939-3945}, publisher = {AAAI}, abstract = {Inspired by recent successes of deep learning in computer vision, we propose a novel framework for encoding time series as different types of images, namely, Gramian Angular Summation/Difference Fields (GASF/GADF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for time series classification and imputation. We used Tiled Convolutional Neural Networks (tiled CNNs) on 20 standard datasets to learn high-level features from the individual and compound GASF-GADF-MTF images. Our approaches achieve highly competitive results when compared to nine of the current best time series classification approaches. Inspired by the bijection property of GASF on 0/1 rescaled data, we train Denoised Auto-encoders (DA) on the GASF images of four standard and one synthesized compound dataset. The imputation MSE on test data is reduced by 12.18%-48.02% when compared to using the raw data. An analysis of the features and weights learned via tiled CNNs and DAs explains why the approaches work }, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Inspired by recent successes of deep learning in computer vision, we propose a novel framework for encoding time series as different types of images, namely, Gramian Angular Summation/Difference Fields (GASF/GADF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for time series classification and imputation. We used Tiled Convolutional Neural Networks (tiled CNNs) on 20 standard datasets to learn high-level features from the individual and compound GASF-GADF-MTF images. Our approaches achieve highly competitive results when compared to nine of the current best time series classification approaches. Inspired by the bijection property of GASF on 0/1 rescaled data, we train Denoised Auto-encoders (DA) on the GASF images of four standard and one synthesized compound dataset. The imputation MSE on test data is reduced by 12.18%-48.02% when compared to using the raw data. An analysis of the features and weights learned via tiled CNNs and DAs explains why the approaches work |
| 5. | Wilkinson, Bryan: Initial Steps for Building a Lexicon of Adjectives with Scalemates. In: Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop, pp. 57–63, Association for Computational Linguistics, Denver, Colorado, 2015. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wilkinson:2015:SRW, title = {Initial Steps for Building a Lexicon of Adjectives with Scalemates}, author = {Bryan Wilkinson}, url = {http://www.aclweb.org/anthology/N15-2008 http://coral-lab.umbc.edu/wp-content/uploads/2015/05/SRW008.pdf}, year = {2015}, date = {2015-06-01}, booktitle = {Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop}, pages = {57--63}, publisher = {Association for Computational Linguistics}, address = {Denver, Colorado}, abstract = {This paper describes work in progress to use clustering to create a lexicon of words that engage in the lexico-semantic relationship known as grading. While other resources like thesauri and taxonomies exist detailing relationships such as synonymy, antonymy, and hyponymy, we do not know of any thorough resource for grading. This work focuses on identifying the words that may participate in this relationship, paving the way for the creation of a true grading lexicon later.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper describes work in progress to use clustering to create a lexicon of words that engage in the lexico-semantic relationship known as grading. While other resources like thesauri and taxonomies exist detailing relationships such as synonymy, antonymy, and hyponymy, we do not know of any thorough resource for grading. This work focuses on identifying the words that may participate in this relationship, paving the way for the creation of a true grading lexicon later. |
| 6. | Wang, Zhiguang; Oates, Tim: Pooling SAX-BoP Approaches with Boosting to Classify Multivariate Synchronous Physiological Time Series Data. In: Proceedings of the Twenty-Eighth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015, Hollywood, Florida. May 18-20, 2015., pp. 335–341, AAAI, 2015. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wang2015pooling, title = {Pooling SAX-BoP Approaches with Boosting to Classify Multivariate Synchronous Physiological Time Series Data}, author = {Zhiguang Wang and Tim Oates}, url = {http://www.aaai.org/ocs/index.php/FLAIRS/FLAIRS15/paper/view/10384}, year = {2015}, date = {2015-05-20}, booktitle = {Proceedings of the Twenty-Eighth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015, Hollywood, Florida. May 18-20, 2015.}, pages = {335--341}, publisher = {AAAI}, crossref = {DBLP:conf/flairs/2015}, abstract = {As the current standard practice of manually recorded vital signs through a few hours is giving way to continuous, automated measurement of high resolution vital signs, it brings a tremendous opportunity to predict patient outcomes and help to improve the early care. However, making predictions in an effective way is fairly challenging, because high resolution vital signs data are multivariate, massive and noisy. Inspired by the max-pooling approaches in Convolutional Neural Networks (CNN), we propose extensions of vanilla SAXBoP approach, called Pooling SAX-BoP to successfully predict patient outcomes from multivariate synchronous vital signs data. Our experiments on two standard datasets demonstrate the Pooling SAX-BoP approaches are competitive with the current state-of-thearts on multivariate time series classification problems. We also integrate Boosting algorithm as one of the most powerful ensemble learning approaches on the BoP representations to further improve the performance. Our experimental results on the clinical data demonstrate that our methods are accurate and stable for classifying multivariate synchronous vital signs time series data.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } As the current standard practice of manually recorded vital signs through a few hours is giving way to continuous, automated measurement of high resolution vital signs, it brings a tremendous opportunity to predict patient outcomes and help to improve the early care. However, making predictions in an effective way is fairly challenging, because high resolution vital signs data are multivariate, massive and noisy. Inspired by the max-pooling approaches in Convolutional Neural Networks (CNN), we propose extensions of vanilla SAXBoP approach, called Pooling SAX-BoP to successfully predict patient outcomes from multivariate synchronous vital signs data. Our experiments on two standard datasets demonstrate the Pooling SAX-BoP approaches are competitive with the current state-of-thearts on multivariate time series classification problems. We also integrate Boosting algorithm as one of the most powerful ensemble learning approaches on the BoP representations to further improve the performance. Our experimental results on the clinical data demonstrate that our methods are accurate and stable for classifying multivariate synchronous vital signs time series data. |
| 7. | Wang, Zhiguang; Oates, Tim: Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks. In: Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence, AAAi, 2015. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wang2015encoding, title = {Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks}, author = {Zhiguang Wang and Tim Oates}, url = {http://coral-lab.umbc.edu/wp-content/uploads/2015/05/10179-43348-1-SM1.pdf}, year = {2015}, date = {2015-01-30}, booktitle = {Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence}, publisher = {AAAi}, abstract = {Inspired by recent successes of deep learning in computer vision and speech recognition, we propose a novel framework to encode time series data as different types of images, namely, Gramian Angular Fields (GAF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for classification. Using a polar coordinate system, GAF images are represented as a Gramian matrix where each element is the trigonometric sum (i.e., superposition of directions) between different time intervals. MTF images represent the first order Markov transition probability along one dimension and temporal dependency along the other. We used Tiled Convolutional Neural Networks (tiled CNNs) on 12 standard datasets to learn high-level features from individual GAF, MTF, and GAF-MTF images that resulted from combining GAF and MTF representations into a single image. The classification results of our approach are competitive with five stateof-the-art approaches. An analysis of the features and weights learned via tiled CNNs explains why the approach works.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Inspired by recent successes of deep learning in computer vision and speech recognition, we propose a novel framework to encode time series data as different types of images, namely, Gramian Angular Fields (GAF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for classification. Using a polar coordinate system, GAF images are represented as a Gramian matrix where each element is the trigonometric sum (i.e., superposition of directions) between different time intervals. MTF images represent the first order Markov transition probability along one dimension and temporal dependency along the other. We used Tiled Convolutional Neural Networks (tiled CNNs) on 12 standard datasets to learn high-level features from individual GAF, MTF, and GAF-MTF images that resulted from combining GAF and MTF representations into a single image. The classification results of our approach are competitive with five stateof-the-art approaches. An analysis of the features and weights learned via tiled CNNs explains why the approach works. |
| 8. | Page, Adam; Sagedy, Chris; Smith, Emily; Attaran, Nasrin; Oates, Tim; Mohsenin, Tinoosh: A Flexible Multichannel EEG Feature Extractor and Classifier for Seizure Detection. In: IEEE Trans. on Circuits and Systems, 62-II (2), pp. 109–113, 2015. (Type: Journal Article | Links | BibTeX) @article{DBLP:journals/tcas/PageSSAOM15, title = {A Flexible Multichannel EEG Feature Extractor and Classifier for Seizure Detection}, author = {Adam Page and Chris Sagedy and Emily Smith and Nasrin Attaran and Tim Oates and Tinoosh Mohsenin}, url = {http://dx.doi.org/10.1109/TCSII.2014.2385211}, doi = {10.1109/TCSII.2014.2385211}, year = {2015}, date = {2015-01-01}, journal = {IEEE Trans. on Circuits and Systems}, volume = {62-II}, number = {2}, pages = {109--113}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 9. | Senin, Pavel; Lin, Jessica; Wang, Xing; Oates, Tim; Gandhi, Sunil; Boedihardjo, Arnold P; Chen, Crystal; Frankenstein, Susan: Time series anomaly discovery with grammar-based compression. In: Proceedings of the 18th International Conference on Extending Database Technology, EDBT 2015, Brussels, Belgium, March 23-27, 2015., pp. 481–492, 2015. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/edbt/Senin0WOGBCF15, title = {Time series anomaly discovery with grammar-based compression}, author = {Pavel Senin and Jessica Lin and Xing Wang and Tim Oates and Sunil Gandhi and Arnold P. Boedihardjo and Crystal Chen and Susan Frankenstein}, url = {http://dx.doi.org/10.5441/002/edbt.2015.42}, doi = {10.5441/002/edbt.2015.42}, year = {2015}, date = {2015-01-01}, booktitle = {Proceedings of the 18th International Conference on Extending Database Technology, EDBT 2015, Brussels, Belgium, March 23-27, 2015.}, pages = {481--492}, crossref = {DBLP:conf/edbt/2015}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 10. | Wang, Zhiguang; Oates, Tim: Time Warping Symbolic Aggregation Approximation with Bag-of-Patterns Representation for Time Series Classification. In: Machine Learning and Applications (ICMLA), 2014 13th International Conference on, pp. 270–275, IEEE, 2014. (Type: Inproceedings | Abstract | Links | BibTeX) @inproceedings{wang2014time, title = {Time Warping Symbolic Aggregation Approximation with Bag-of-Patterns Representation for Time Series Classification}, author = {Zhiguang Wang and Tim Oates}, url = {http://coral-lab.umbc.edu/wp-content/uploads/2015/05/Wang.pdf}, doi = {10.1109/ICMLA.2014.49}, year = {2014}, date = {2014-12-03}, booktitle = {Machine Learning and Applications (ICMLA), 2014 13th International Conference on}, pages = {270--275}, publisher = {IEEE}, abstract = {Standard Symbolic Aggregation Approximation (SAX) is at the core of many effective time series data mining algorithms. Its combination with Bag-of-Patterns (BoP) has become the standard approach with state-of-the-art performance on standard datasets. However, standard SAX with the BoP representation might neglect internal temporal correlation embedded in the raw data. In this paper, we proposed time warping SAX, which extends the standard SAX with time delay embedding vector approaches to account for temporal correlations. We test time warping SAX with the BoP representation on 12 benchmark datasets from the UCR Time Series Classification/Clustering Collection. On 9 datasets, time warping SAX overtakes the state-of-the-art performance of the standard SAX. To validate our methods in real world applications, a new dataset of vital signs data collected from patients who may require blood transfusion in the next 6 hours was tested. All the results demonstrate that, by considering the temporal internal correlation, time warping SAX combined with BoP improves classification performance.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Standard Symbolic Aggregation Approximation (SAX) is at the core of many effective time series data mining algorithms. Its combination with Bag-of-Patterns (BoP) has become the standard approach with state-of-the-art performance on standard datasets. However, standard SAX with the BoP representation might neglect internal temporal correlation embedded in the raw data. In this paper, we proposed time warping SAX, which extends the standard SAX with time delay embedding vector approaches to account for temporal correlations. We test time warping SAX with the BoP representation on 12 benchmark datasets from the UCR Time Series Classification/Clustering Collection. On 9 datasets, time warping SAX overtakes the state-of-the-art performance of the standard SAX. To validate our methods in real world applications, a new dataset of vital signs data collected from patients who may require blood transfusion in the next 6 hours was tested. All the results demonstrate that, by considering the temporal internal correlation, time warping SAX combined with BoP improves classification performance. |
| 11. | Zhu, Xianshu; Oates, Tim: Finding story chains in newswire articles using random walks. In: Information Systems Frontiers, 16 (5), pp. 753–769, 2014. (Type: Journal Article | Links | BibTeX) @article{DBLP:journals/isf/ZhuO14, title = {Finding story chains in newswire articles using random walks}, author = {Xianshu Zhu and Tim Oates}, url = {http://dx.doi.org/10.1007/s10796-013-9420-2}, doi = {10.1007/s10796-013-9420-2}, year = {2014}, date = {2014-01-01}, journal = {Information Systems Frontiers}, volume = {16}, number = {5}, pages = {753--769}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 12. | Heinz, Jeffrey; de la Higuera, Colin; Oates, Tim: Introduction to the Special Issue on Grammatical Inference. In: Machine Learning, 96 (1-2), pp. 1–3, 2014. (Type: Journal Article | Links | BibTeX) @article{DBLP:journals/ml/HeinzHO14, title = {Introduction to the Special Issue on Grammatical Inference}, author = {Jeffrey Heinz and Colin de la Higuera and Tim Oates}, url = {http://dx.doi.org/10.1007/s10994-013-5428-6}, doi = {10.1007/s10994-013-5428-6}, year = {2014}, date = {2014-01-01}, journal = {Machine Learning}, volume = {96}, number = {1-2}, pages = {1--3}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 13. | Page, Adam; Turner, J T; Mohsenin, Tinoosh; Oates, Tim: Comparing Raw Data and Feature Extraction for Seizure Detection with Deep Learning Methods. In: Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society Conference, FLAIRS 2014, Pensacola Beach, Florida, May 21-23, 2014., 2014. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/flairs/PageTMO14, title = {Comparing Raw Data and Feature Extraction for Seizure Detection with Deep Learning Methods}, author = {Adam Page and J. T. Turner and Tinoosh Mohsenin and Tim Oates}, url = {http://www.aaai.org/ocs/index.php/FLAIRS/FLAIRS14/paper/view/7885}, year = {2014}, date = {2014-01-01}, booktitle = {Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society Conference, FLAIRS 2014, Pensacola Beach, Florida, May 21-23, 2014.}, crossref = {DBLP:conf/flairs/2014}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 14. | Deivachilai, Rakesh; Oates, Tim: On-Line Signature Verification Using Symbolic Aggregate Approximation (SAX) and Sequential Mining Optimization (SMO). In: 13th International Conference on Machine Learning and Applications, ICMLA 2014, Detroit, MI, USA, December 3-6, 2014, pp. 195–200, 2014. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/icmla/DeivachilaiO14, title = {On-Line Signature Verification Using Symbolic Aggregate Approximation (SAX) and Sequential Mining Optimization (SMO)}, author = {Rakesh Deivachilai and Tim Oates}, url = {http://dx.doi.org/10.1109/ICMLA.2014.36}, doi = {10.1109/ICMLA.2014.36}, year = {2014}, date = {2014-01-01}, booktitle = {13th International Conference on Machine Learning and Applications, ICMLA 2014, Detroit, MI, USA, December 3-6, 2014}, pages = {195--200}, crossref = {DBLP:conf/icmla/2014}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 15. | Senin, Pavel; Lin, Jessica; Wang, Xing; Oates, Tim; Gandhi, Sunil; Boedihardjo, Arnold P; Chen, Crystal; Frankenstein, Susan; Lerner, Manfred: GrammarViz 2.0: A Tool for Grammar-Based Pattern Discovery in Time Series. In: Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2014, Nancy, France, September 15-19, 2014. Proceedings, Part III, pp. 468–472, 2014. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/pkdd/Senin0WOGBCFL14, title = {GrammarViz 2.0: A Tool for Grammar-Based Pattern Discovery in Time Series}, author = {Pavel Senin and Jessica Lin and Xing Wang and Tim Oates and Sunil Gandhi and Arnold P. Boedihardjo and Crystal Chen and Susan Frankenstein and Manfred Lerner}, url = {http://dx.doi.org/10.1007/978-3-662-44845-8_37}, doi = {10.1007/978-3-662-44845-8_37}, year = {2014}, date = {2014-01-01}, booktitle = {Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2014, Nancy, France, September 15-19, 2014. Proceedings, Part III}, pages = {468--472}, crossref = {DBLP:conf/pkdd/2014-3}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 16. | Pramod, Siddharth; Page, Adam; Mohsenin, Tinoosh; Oates, Tim: Detecting Epileptic Seizures from EEG Data using Neural Networks. In: CoRR, abs/1412.6502 , 2014. (Type: Journal Article | Links | BibTeX) @article{DBLP:journals/corr/PramodPMO14, title = {Detecting Epileptic Seizures from EEG Data using Neural Networks}, author = {Siddharth Pramod and Adam Page and Tinoosh Mohsenin and Tim Oates}, url = {http://arxiv.org/abs/1412.6502}, year = {2014}, date = {2014-01-01}, journal = {CoRR}, volume = {abs/1412.6502}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 17. | Tsai, Terry H; Kasch, Niels; Pfeifer, Craig; Oates, Tim: Text Mining for Hypotheses and Results in Translational Medicine Studies. In: 2014 IEEE International Conference on Data Mining Workshops, ICDM Workshops 2014, Shenzhen, China, December 14, 2014, pp. 127–132, 2014. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/icdm/TsaiKPO14b, title = {Text Mining for Hypotheses and Results in Translational Medicine Studies}, author = { Terry H. Tsai and Niels Kasch and Craig Pfeifer and Tim Oates}, url = {http://dx.doi.org/10.1109/ICDMW.2014.39}, doi = {10.1109/ICDMW.2014.39}, year = {2014}, date = {2014-01-01}, booktitle = {2014 IEEE International Conference on Data Mining Workshops, ICDM Workshops 2014, Shenzhen, China, December 14, 2014}, pages = {127--132}, crossref = {DBLP:conf/icdm/2014w}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 18. | Tim Oates Arnold P. Boedihardjo, Jessica Lin Crystal Chen Susan Frankenstein Sunil Gandhi : Motif Discovery in Spatial Trajectories Using Grammar Inference. Proceedings of the 22Nd ACM International Conference on Conference on Information Knowledge Management
, ACM, 2013, ISBN: 978-1-4503-2263-8. (Type: Conference | Abstract | Links | BibTeX) @conference{Oates:2013:MDS:2505515.2507820, title = {Motif Discovery in Spatial Trajectories Using Grammar Inference}, author = {Tim Oates, Arnold P. Boedihardjo, Jessica Lin, Crystal Chen, Susan Frankenstein, Sunil Gandhi}, url = {http://coral-lab.umbc.edu/wp-content/uploads/2015/05/p1465-oates.pdf}, doi = {10.1145/2505515.2507820}, isbn = {978-1-4503-2263-8}, year = {2013}, date = {2013-05-01}, booktitle = {Proceedings of the 22Nd ACM International Conference on Conference on Information Knowledge Management }, issuetitle = {Motif Discovery in Spatial Trajectories Using Grammar Inference}, publisher = {ACM}, abstract = {Spatial trajectory analysis is crucial to uncovering insights into the motives and nature of human behavior. In this work, we study the problem of discovering motifs in trajectories based on symbolically transformed representations and context free grammars. We propose a fast and robust grammar induction algorithm called mSEQUITUR to infer a grammar rule set from a trajectory for motif generation. Second, we designed the Symbolic Trajectory Analysis and VIsualization System (STAVIS), the first of its kind trajectory analytical system that applies grammar inference to derive trajectory signatures and enable mining tasks on the signatures. Third, an empirical evaluation is performed to demonstrate the efficiency and effectiveness of mSEQUITUR for generating trajectory signatures and discovering motifs.}, keywords = {}, pubstate = {published}, tppubtype = {conference} } Spatial trajectory analysis is crucial to uncovering insights into the motives and nature of human behavior. In this work, we study the problem of discovering motifs in trajectories based on symbolically transformed representations and context free grammars. We propose a fast and robust grammar induction algorithm called mSEQUITUR to infer a grammar rule set from a trajectory for motif generation. Second, we designed the Symbolic Trajectory Analysis and VIsualization System (STAVIS), the first of its kind trajectory analytical system that applies grammar inference to derive trajectory signatures and enable mining tasks on the signatures. Third, an empirical evaluation is performed to demonstrate the efficiency and effectiveness of mSEQUITUR for generating trajectory signatures and discovering motifs. |
| 19. | Oates, Tim; Boedihardjo, Arnold P; Lin, Jessica; Chen, Crystal; Frankenstein, Susan; Gandhi, Sunil: Motif discovery in spatial trajectories using grammar inference. In: 22nd ACM International Conference on Information and Knowledge Management, CIKM'13, San Francisco, CA, USA, October 27 - November 1, 2013, pp. 1465–1468, 2013. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/cikm/OatesB0CFG13, title = {Motif discovery in spatial trajectories using grammar inference}, author = {Tim Oates and Arnold P. Boedihardjo and Jessica Lin and Crystal Chen and Susan Frankenstein and Sunil Gandhi}, url = {http://doi.acm.org/10.1145/2505515.2507820}, doi = {10.1145/2505515.2507820}, year = {2013}, date = {2013-01-01}, booktitle = {22nd ACM International Conference on Information and Knowledge Management, CIKM'13, San Francisco, CA, USA, October 27 - November 1, 2013}, pages = {1465--1468}, crossref = {DBLP:conf/cikm/2013}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 20. | Rosebrock, Adrian; Oates, Tim; Caban, Jesus: Ecosembles: A Rapidly Deployable Image Classification System Using Feature-Views. In: 12th International Conference on Machine Learning and Applications, ICMLA 2013, Miami, FL, USA, December 4-7, 2013, Volume 1, pp. 1–8, 2013. (Type: Inproceedings | Links | BibTeX) @inproceedings{DBLP:conf/icmla/RosebrockOC13, title = {Ecosembles: A Rapidly Deployable Image Classification System Using Feature-Views}, author = {Adrian Rosebrock and Tim Oates and Jesus Caban}, url = {http://dx.doi.org/10.1109/ICMLA.2013.9}, doi = {10.1109/ICMLA.2013.9}, year = {2013}, date = {2013-01-01}, booktitle = {12th International Conference on Machine Learning and Applications, ICMLA 2013, Miami, FL, USA, December 4-7, 2013, Volume 1}, pages = {1--8}, crossref = {DBLP:conf/icmla/2013-1}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
2016 |
Wilkinson, Bryan; Oates, Tim A Gold Standard for Scalar Adjectives Conference Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016), European Language Resources Association (ELRA), Portorož, Slovenia, 2016. @conference{Wilkinson2016, title = {A Gold Standard for Scalar Adjectives}, author = {Bryan Wilkinson and Tim Oates }, url = {http://www.lrec-conf.org/proceedings/lrec2016/pdf/1004_Paper.pdf}, year = {2016}, date = {2016-05-23}, booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016)}, publisher = {European Language Resources Association (ELRA)}, address = {Portorož, Slovenia}, keywords = {}, pubstate = {published}, tppubtype = {conference} } |
Zhiguang Wang Tim Oates, James Lo Adaptive Normalized Risk-Averting Training for Deep Neural Networks Inproceedings Proceedings of The Thirtieth AAAI Conference on Artificial Intelligence, AAAI, 2016. @inproceedings{wang2016adaptive, title = {Adaptive Normalized Risk-Averting Training for Deep Neural Networks}, author = {Zhiguang Wang, Tim Oates, James Lo}, url = {http://www.aaai.org/ocs/index.php/AAAI/AAAI16/paper/view/11765}, year = {2016}, date = {2016-02-17}, booktitle = {Proceedings of The Thirtieth AAAI Conference on Artificial Intelligence}, publisher = {AAAI}, abstract = {This paper proposes a set of new error criteria and a learning approach, called Adaptive Normalized Risk-Averting Training (ANRAT) to attack the non-convex optimization problem in training deep neural networks without pretraining. Theoretically, we demonstrate its effectiveness based on the expansion of the convexity region. By analyzing the gradient on the convexity index $\lambda$, we explain the reason why our learning method using gradient descent works. In practice, we show how this training method is successfully applied for improved training of deep neural networks to solve visual recognition tasks on the MNIST and CIFAR-10 datasets. Using simple experimental settings without pretraining and other tricks, we obtain results comparable or superior to those reported in recent literature on the same tasks using standard ConvNets + MSE/cross entropy. Performance on deep/shallow multilayer perceptron and Denoised Auto-encoder is also explored. ANRAT can be combined with other quasi-Newton training methods, innovative network variants, regularization techniques and other common tricks in DNNs. Other than unsupervised pretraining, it provides a new perspective to address the non-convex optimization strategy in training DNNs.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper proposes a set of new error criteria and a learning approach, called Adaptive Normalized Risk-Averting Training (ANRAT) to attack the non-convex optimization problem in training deep neural networks without pretraining. Theoretically, we demonstrate its effectiveness based on the expansion of the convexity region. By analyzing the gradient on the convexity index $lambda$, we explain the reason why our learning method using gradient descent works. In practice, we show how this training method is successfully applied for improved training of deep neural networks to solve visual recognition tasks on the MNIST and CIFAR-10 datasets. Using simple experimental settings without pretraining and other tricks, we obtain results comparable or superior to those reported in recent literature on the same tasks using standard ConvNets + MSE/cross entropy. Performance on deep/shallow multilayer perceptron and Denoised Auto-encoder is also explored. ANRAT can be combined with other quasi-Newton training methods, innovative network variants, regularization techniques and other common tricks in DNNs. Other than unsupervised pretraining, it provides a new perspective to address the non-convex optimization strategy in training DNNs. |
2015 |
Clemens, John Automatic Classification of Object Code using Machine Learning Inproceedings Digital Investigation , pp. S156 - S162, 2015, ISSN: 1742-2876, (DFRWS USA 2015). @inproceedings{Clemens2015, title = {Automatic Classification of Object Code using Machine Learning}, author = {John Clemens}, url = {http://www.sciencedirect.com/science/article/pii/S1742287615000523 http://www.sciencedirect.com/science/article/pii/S1742287615000523/pdfft?md5=a270d4d9b1816b67a51e59ba3e6e2284&pid=1-s2.0-S1742287615000523-main.pdf}, doi = {10.1016/j.diin.2015.05.007}, issn = {1742-2876}, year = {2015}, date = {2015-08-09}, urldate = {2015-10-08}, booktitle = {Digital Investigation }, issuetitle = {The Proceedings of the Fifteenth Annual DFRWS Conference }, journal = {Digital Investigation }, volume = {14, Supplement 1}, pages = {S156 - S162}, abstract = {Abstract Recent research has repeatedly shown that machine learning techniques can be applied to either whole files or file fragments to classify them for analysis. We build upon these techniques to show that for samples of un-labeled compiled computer object code, one can apply the same type of analysis to classify important aspects of the code, such as its target architecture and endianess. We show that using simple byte-value histograms we retain enough information about the opcodes within a sample to classify the target architecture with high accuracy, and then discuss heuristic-based features that exploit information within the operands to determine endianess. We introduce a dataset with over 16000 code samples from 20 architectures and experimentally show that by using our features, classifiers can achieve very high accuracy with relatively small sample sizes. }, note = {DFRWS USA 2015}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Abstract Recent research has repeatedly shown that machine learning techniques can be applied to either whole files or file fragments to classify them for analysis. We build upon these techniques to show that for samples of un-labeled compiled computer object code, one can apply the same type of analysis to classify important aspects of the code, such as its target architecture and endianess. We show that using simple byte-value histograms we retain enough information about the opcodes within a sample to classify the target architecture with high accuracy, and then discuss heuristic-based features that exploit information within the operands to determine endianess. We introduce a dataset with over 16000 code samples from 20 architectures and experimentally show that by using our features, classifiers can achieve very high accuracy with relatively small sample sizes. |
Zhiguang wang, Tim Oates Imaging Time-Series to Improve Classification and Imputation Inproceedings Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI), pp. 3939-3945, AAAI, 2015. @inproceedings{wang2015imaging, title = {Imaging Time-Series to Improve Classification and Imputation}, author = {Zhiguang wang, Tim Oates}, url = {http://ijcai.org/papers15/Papers/IJCAI15-553.pdf}, year = {2015}, date = {2015-08-01}, booktitle = {Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI)}, pages = {3939-3945}, publisher = {AAAI}, abstract = {Inspired by recent successes of deep learning in computer vision, we propose a novel framework for encoding time series as different types of images, namely, Gramian Angular Summation/Difference Fields (GASF/GADF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for time series classification and imputation. We used Tiled Convolutional Neural Networks (tiled CNNs) on 20 standard datasets to learn high-level features from the individual and compound GASF-GADF-MTF images. Our approaches achieve highly competitive results when compared to nine of the current best time series classification approaches. Inspired by the bijection property of GASF on 0/1 rescaled data, we train Denoised Auto-encoders (DA) on the GASF images of four standard and one synthesized compound dataset. The imputation MSE on test data is reduced by 12.18%-48.02% when compared to using the raw data. An analysis of the features and weights learned via tiled CNNs and DAs explains why the approaches work }, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Inspired by recent successes of deep learning in computer vision, we propose a novel framework for encoding time series as different types of images, namely, Gramian Angular Summation/Difference Fields (GASF/GADF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for time series classification and imputation. We used Tiled Convolutional Neural Networks (tiled CNNs) on 20 standard datasets to learn high-level features from the individual and compound GASF-GADF-MTF images. Our approaches achieve highly competitive results when compared to nine of the current best time series classification approaches. Inspired by the bijection property of GASF on 0/1 rescaled data, we train Denoised Auto-encoders (DA) on the GASF images of four standard and one synthesized compound dataset. The imputation MSE on test data is reduced by 12.18%-48.02% when compared to using the raw data. An analysis of the features and weights learned via tiled CNNs and DAs explains why the approaches work |
Wilkinson, Bryan Initial Steps for Building a Lexicon of Adjectives with Scalemates Inproceedings Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop, pp. 57–63, Association for Computational Linguistics, Denver, Colorado, 2015. @inproceedings{wilkinson:2015:SRW, title = {Initial Steps for Building a Lexicon of Adjectives with Scalemates}, author = {Bryan Wilkinson}, url = {http://www.aclweb.org/anthology/N15-2008 http://coral-lab.umbc.edu/wp-content/uploads/2015/05/SRW008.pdf}, year = {2015}, date = {2015-06-01}, booktitle = {Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop}, pages = {57--63}, publisher = {Association for Computational Linguistics}, address = {Denver, Colorado}, abstract = {This paper describes work in progress to use clustering to create a lexicon of words that engage in the lexico-semantic relationship known as grading. While other resources like thesauri and taxonomies exist detailing relationships such as synonymy, antonymy, and hyponymy, we do not know of any thorough resource for grading. This work focuses on identifying the words that may participate in this relationship, paving the way for the creation of a true grading lexicon later.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } This paper describes work in progress to use clustering to create a lexicon of words that engage in the lexico-semantic relationship known as grading. While other resources like thesauri and taxonomies exist detailing relationships such as synonymy, antonymy, and hyponymy, we do not know of any thorough resource for grading. This work focuses on identifying the words that may participate in this relationship, paving the way for the creation of a true grading lexicon later. |
Wang, Zhiguang; Oates, Tim Pooling SAX-BoP Approaches with Boosting to Classify Multivariate Synchronous Physiological Time Series Data Inproceedings Proceedings of the Twenty-Eighth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015, Hollywood, Florida. May 18-20, 2015., pp. 335–341, AAAI, 2015. @inproceedings{wang2015pooling, title = {Pooling SAX-BoP Approaches with Boosting to Classify Multivariate Synchronous Physiological Time Series Data}, author = {Zhiguang Wang and Tim Oates}, url = {http://www.aaai.org/ocs/index.php/FLAIRS/FLAIRS15/paper/view/10384}, year = {2015}, date = {2015-05-20}, booktitle = {Proceedings of the Twenty-Eighth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015, Hollywood, Florida. May 18-20, 2015.}, pages = {335--341}, publisher = {AAAI}, crossref = {DBLP:conf/flairs/2015}, abstract = {As the current standard practice of manually recorded vital signs through a few hours is giving way to continuous, automated measurement of high resolution vital signs, it brings a tremendous opportunity to predict patient outcomes and help to improve the early care. However, making predictions in an effective way is fairly challenging, because high resolution vital signs data are multivariate, massive and noisy. Inspired by the max-pooling approaches in Convolutional Neural Networks (CNN), we propose extensions of vanilla SAXBoP approach, called Pooling SAX-BoP to successfully predict patient outcomes from multivariate synchronous vital signs data. Our experiments on two standard datasets demonstrate the Pooling SAX-BoP approaches are competitive with the current state-of-thearts on multivariate time series classification problems. We also integrate Boosting algorithm as one of the most powerful ensemble learning approaches on the BoP representations to further improve the performance. Our experimental results on the clinical data demonstrate that our methods are accurate and stable for classifying multivariate synchronous vital signs time series data.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } As the current standard practice of manually recorded vital signs through a few hours is giving way to continuous, automated measurement of high resolution vital signs, it brings a tremendous opportunity to predict patient outcomes and help to improve the early care. However, making predictions in an effective way is fairly challenging, because high resolution vital signs data are multivariate, massive and noisy. Inspired by the max-pooling approaches in Convolutional Neural Networks (CNN), we propose extensions of vanilla SAXBoP approach, called Pooling SAX-BoP to successfully predict patient outcomes from multivariate synchronous vital signs data. Our experiments on two standard datasets demonstrate the Pooling SAX-BoP approaches are competitive with the current state-of-thearts on multivariate time series classification problems. We also integrate Boosting algorithm as one of the most powerful ensemble learning approaches on the BoP representations to further improve the performance. Our experimental results on the clinical data demonstrate that our methods are accurate and stable for classifying multivariate synchronous vital signs time series data. |
Wang, Zhiguang; Oates, Tim Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks Inproceedings Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence, AAAi, 2015. @inproceedings{wang2015encoding, title = {Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks}, author = {Zhiguang Wang and Tim Oates}, url = {http://coral-lab.umbc.edu/wp-content/uploads/2015/05/10179-43348-1-SM1.pdf}, year = {2015}, date = {2015-01-30}, booktitle = {Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence}, publisher = {AAAi}, abstract = {Inspired by recent successes of deep learning in computer vision and speech recognition, we propose a novel framework to encode time series data as different types of images, namely, Gramian Angular Fields (GAF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for classification. Using a polar coordinate system, GAF images are represented as a Gramian matrix where each element is the trigonometric sum (i.e., superposition of directions) between different time intervals. MTF images represent the first order Markov transition probability along one dimension and temporal dependency along the other. We used Tiled Convolutional Neural Networks (tiled CNNs) on 12 standard datasets to learn high-level features from individual GAF, MTF, and GAF-MTF images that resulted from combining GAF and MTF representations into a single image. The classification results of our approach are competitive with five stateof-the-art approaches. An analysis of the features and weights learned via tiled CNNs explains why the approach works.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Inspired by recent successes of deep learning in computer vision and speech recognition, we propose a novel framework to encode time series data as different types of images, namely, Gramian Angular Fields (GAF) and Markov Transition Fields (MTF). This enables the use of techniques from computer vision for classification. Using a polar coordinate system, GAF images are represented as a Gramian matrix where each element is the trigonometric sum (i.e., superposition of directions) between different time intervals. MTF images represent the first order Markov transition probability along one dimension and temporal dependency along the other. We used Tiled Convolutional Neural Networks (tiled CNNs) on 12 standard datasets to learn high-level features from individual GAF, MTF, and GAF-MTF images that resulted from combining GAF and MTF representations into a single image. The classification results of our approach are competitive with five stateof-the-art approaches. An analysis of the features and weights learned via tiled CNNs explains why the approach works. |
Page, Adam; Sagedy, Chris; Smith, Emily; Attaran, Nasrin; Oates, Tim; Mohsenin, Tinoosh A Flexible Multichannel EEG Feature Extractor and Classifier for Seizure Detection Journal Article IEEE Trans. on Circuits and Systems, 62-II (2), pp. 109–113, 2015. @article{DBLP:journals/tcas/PageSSAOM15, title = {A Flexible Multichannel EEG Feature Extractor and Classifier for Seizure Detection}, author = {Adam Page and Chris Sagedy and Emily Smith and Nasrin Attaran and Tim Oates and Tinoosh Mohsenin}, url = {http://dx.doi.org/10.1109/TCSII.2014.2385211}, doi = {10.1109/TCSII.2014.2385211}, year = {2015}, date = {2015-01-01}, journal = {IEEE Trans. on Circuits and Systems}, volume = {62-II}, number = {2}, pages = {109--113}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Senin, Pavel; Lin, Jessica; Wang, Xing; Oates, Tim; Gandhi, Sunil; Boedihardjo, Arnold P; Chen, Crystal; Frankenstein, Susan Time series anomaly discovery with grammar-based compression Inproceedings Proceedings of the 18th International Conference on Extending Database Technology, EDBT 2015, Brussels, Belgium, March 23-27, 2015., pp. 481–492, 2015. @inproceedings{DBLP:conf/edbt/Senin0WOGBCF15, title = {Time series anomaly discovery with grammar-based compression}, author = {Pavel Senin and Jessica Lin and Xing Wang and Tim Oates and Sunil Gandhi and Arnold P. Boedihardjo and Crystal Chen and Susan Frankenstein}, url = {http://dx.doi.org/10.5441/002/edbt.2015.42}, doi = {10.5441/002/edbt.2015.42}, year = {2015}, date = {2015-01-01}, booktitle = {Proceedings of the 18th International Conference on Extending Database Technology, EDBT 2015, Brussels, Belgium, March 23-27, 2015.}, pages = {481--492}, crossref = {DBLP:conf/edbt/2015}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
2014 |
Wang, Zhiguang; Oates, Tim Time Warping Symbolic Aggregation Approximation with Bag-of-Patterns Representation for Time Series Classification Inproceedings Machine Learning and Applications (ICMLA), 2014 13th International Conference on, pp. 270–275, IEEE, 2014. @inproceedings{wang2014time, title = {Time Warping Symbolic Aggregation Approximation with Bag-of-Patterns Representation for Time Series Classification}, author = {Zhiguang Wang and Tim Oates}, url = {http://coral-lab.umbc.edu/wp-content/uploads/2015/05/Wang.pdf}, doi = {10.1109/ICMLA.2014.49}, year = {2014}, date = {2014-12-03}, booktitle = {Machine Learning and Applications (ICMLA), 2014 13th International Conference on}, pages = {270--275}, publisher = {IEEE}, abstract = {Standard Symbolic Aggregation Approximation (SAX) is at the core of many effective time series data mining algorithms. Its combination with Bag-of-Patterns (BoP) has become the standard approach with state-of-the-art performance on standard datasets. However, standard SAX with the BoP representation might neglect internal temporal correlation embedded in the raw data. In this paper, we proposed time warping SAX, which extends the standard SAX with time delay embedding vector approaches to account for temporal correlations. We test time warping SAX with the BoP representation on 12 benchmark datasets from the UCR Time Series Classification/Clustering Collection. On 9 datasets, time warping SAX overtakes the state-of-the-art performance of the standard SAX. To validate our methods in real world applications, a new dataset of vital signs data collected from patients who may require blood transfusion in the next 6 hours was tested. All the results demonstrate that, by considering the temporal internal correlation, time warping SAX combined with BoP improves classification performance.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Standard Symbolic Aggregation Approximation (SAX) is at the core of many effective time series data mining algorithms. Its combination with Bag-of-Patterns (BoP) has become the standard approach with state-of-the-art performance on standard datasets. However, standard SAX with the BoP representation might neglect internal temporal correlation embedded in the raw data. In this paper, we proposed time warping SAX, which extends the standard SAX with time delay embedding vector approaches to account for temporal correlations. We test time warping SAX with the BoP representation on 12 benchmark datasets from the UCR Time Series Classification/Clustering Collection. On 9 datasets, time warping SAX overtakes the state-of-the-art performance of the standard SAX. To validate our methods in real world applications, a new dataset of vital signs data collected from patients who may require blood transfusion in the next 6 hours was tested. All the results demonstrate that, by considering the temporal internal correlation, time warping SAX combined with BoP improves classification performance. |
Zhu, Xianshu; Oates, Tim Finding story chains in newswire articles using random walks Journal Article Information Systems Frontiers, 16 (5), pp. 753–769, 2014. @article{DBLP:journals/isf/ZhuO14, title = {Finding story chains in newswire articles using random walks}, author = {Xianshu Zhu and Tim Oates}, url = {http://dx.doi.org/10.1007/s10796-013-9420-2}, doi = {10.1007/s10796-013-9420-2}, year = {2014}, date = {2014-01-01}, journal = {Information Systems Frontiers}, volume = {16}, number = {5}, pages = {753--769}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Heinz, Jeffrey; de la Higuera, Colin; Oates, Tim Introduction to the Special Issue on Grammatical Inference Journal Article Machine Learning, 96 (1-2), pp. 1–3, 2014. @article{DBLP:journals/ml/HeinzHO14, title = {Introduction to the Special Issue on Grammatical Inference}, author = {Jeffrey Heinz and Colin de la Higuera and Tim Oates}, url = {http://dx.doi.org/10.1007/s10994-013-5428-6}, doi = {10.1007/s10994-013-5428-6}, year = {2014}, date = {2014-01-01}, journal = {Machine Learning}, volume = {96}, number = {1-2}, pages = {1--3}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Page, Adam; Turner, J T; Mohsenin, Tinoosh; Oates, Tim Comparing Raw Data and Feature Extraction for Seizure Detection with Deep Learning Methods Inproceedings Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society Conference, FLAIRS 2014, Pensacola Beach, Florida, May 21-23, 2014., 2014. @inproceedings{DBLP:conf/flairs/PageTMO14, title = {Comparing Raw Data and Feature Extraction for Seizure Detection with Deep Learning Methods}, author = {Adam Page and J. T. Turner and Tinoosh Mohsenin and Tim Oates}, url = {http://www.aaai.org/ocs/index.php/FLAIRS/FLAIRS14/paper/view/7885}, year = {2014}, date = {2014-01-01}, booktitle = {Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society Conference, FLAIRS 2014, Pensacola Beach, Florida, May 21-23, 2014.}, crossref = {DBLP:conf/flairs/2014}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
Deivachilai, Rakesh; Oates, Tim On-Line Signature Verification Using Symbolic Aggregate Approximation (SAX) and Sequential Mining Optimization (SMO) Inproceedings 13th International Conference on Machine Learning and Applications, ICMLA 2014, Detroit, MI, USA, December 3-6, 2014, pp. 195–200, 2014. @inproceedings{DBLP:conf/icmla/DeivachilaiO14, title = {On-Line Signature Verification Using Symbolic Aggregate Approximation (SAX) and Sequential Mining Optimization (SMO)}, author = {Rakesh Deivachilai and Tim Oates}, url = {http://dx.doi.org/10.1109/ICMLA.2014.36}, doi = {10.1109/ICMLA.2014.36}, year = {2014}, date = {2014-01-01}, booktitle = {13th International Conference on Machine Learning and Applications, ICMLA 2014, Detroit, MI, USA, December 3-6, 2014}, pages = {195--200}, crossref = {DBLP:conf/icmla/2014}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
Senin, Pavel; Lin, Jessica; Wang, Xing; Oates, Tim; Gandhi, Sunil; Boedihardjo, Arnold P; Chen, Crystal; Frankenstein, Susan; Lerner, Manfred GrammarViz 2.0: A Tool for Grammar-Based Pattern Discovery in Time Series Inproceedings Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2014, Nancy, France, September 15-19, 2014. Proceedings, Part III, pp. 468–472, 2014. @inproceedings{DBLP:conf/pkdd/Senin0WOGBCFL14, title = {GrammarViz 2.0: A Tool for Grammar-Based Pattern Discovery in Time Series}, author = {Pavel Senin and Jessica Lin and Xing Wang and Tim Oates and Sunil Gandhi and Arnold P. Boedihardjo and Crystal Chen and Susan Frankenstein and Manfred Lerner}, url = {http://dx.doi.org/10.1007/978-3-662-44845-8_37}, doi = {10.1007/978-3-662-44845-8_37}, year = {2014}, date = {2014-01-01}, booktitle = {Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2014, Nancy, France, September 15-19, 2014. Proceedings, Part III}, pages = {468--472}, crossref = {DBLP:conf/pkdd/2014-3}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |