Margin-maximizing classification of sequential data with infinitely-long temporal dependencies
► We present a method for sequential data modeling. ► Our approach models temporal dependencies of infinite length. ► It employs a margin maximization training scheme. ► We evaluate it in computer vision applications. Generative models for sequential data are usually based on the assumption of tempo...
Saved in:
| Published in: | Expert systems with applications Vol. 40; no. 11; pp. 4519 - 4527 |
|---|---|
| Main Author: | |
| Format: | Journal Article |
| Language: | English |
| Published: |
Amsterdam
Elsevier Ltd
01-09-2013
Elsevier |
| Subjects: | |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Abstract | ► We present a method for sequential data modeling. ► Our approach models temporal dependencies of infinite length. ► It employs a margin maximization training scheme. ► We evaluate it in computer vision applications.
Generative models for sequential data are usually based on the assumption of temporal dependencies described by a first-order Markov chain. To ameliorate this shallow modeling assumption, several authors have proposed models with higher-order dependencies. However, the practical applicability of these approaches is hindered by their prohibitive computational costs in most cases. In addition, most existing approaches give rise to model training algorithms with objective functions that entail multiple spurious local optima, thus requiring application of tedious countermeasures to avoid getting trapped to bad model estimates. In this paper, we devise a novel margin-maximizing model with convex objective function that allows for capturing infinitely-long temporal dependencies in sequential datasets. This is effected by utilizing a recently proposed nonparametric Bayesian model of label sequences with infinitely-long temporal dependencies, namely the sequence memoizer, and training our model using margin maximization and a versatile mean-field-like approximation to allow for increased computational efficiency. As we experimentally demonstrate, the devised margin-maximizing construction of our model, which leads to a convex optimization scheme, without any spurious local optima, combined with the capacity of our model to capture long and complex temporal dependencies, allow for obtaining exceptional pattern recognition performance in several applications. |
|---|---|
| AbstractList | ► We present a method for sequential data modeling. ► Our approach models temporal dependencies of infinite length. ► It employs a margin maximization training scheme. ► We evaluate it in computer vision applications.
Generative models for sequential data are usually based on the assumption of temporal dependencies described by a first-order Markov chain. To ameliorate this shallow modeling assumption, several authors have proposed models with higher-order dependencies. However, the practical applicability of these approaches is hindered by their prohibitive computational costs in most cases. In addition, most existing approaches give rise to model training algorithms with objective functions that entail multiple spurious local optima, thus requiring application of tedious countermeasures to avoid getting trapped to bad model estimates. In this paper, we devise a novel margin-maximizing model with convex objective function that allows for capturing infinitely-long temporal dependencies in sequential datasets. This is effected by utilizing a recently proposed nonparametric Bayesian model of label sequences with infinitely-long temporal dependencies, namely the sequence memoizer, and training our model using margin maximization and a versatile mean-field-like approximation to allow for increased computational efficiency. As we experimentally demonstrate, the devised margin-maximizing construction of our model, which leads to a convex optimization scheme, without any spurious local optima, combined with the capacity of our model to capture long and complex temporal dependencies, allow for obtaining exceptional pattern recognition performance in several applications. Generative models for sequential data are usually based on the assumption of temporal dependencies described by a first-order Markov chain. To ameliorate this shallow modeling assumption, several authors have proposed models with higher-order dependencies. However, the practical applicability of these approaches is hindered by their prohibitive computational costs in most cases. In addition, most existing approaches give rise to model training algorithms with objective functions that entail multiple spurious local optima, thus requiring application of tedious countermeasures to avoid getting trapped to bad model estimates. In this paper, we devise a novel margin-maximizing model with convex objective function that allows for capturing infinitely-long temporal dependencies in sequential datasets. This is effected by utilizing a recently proposed nonparametric Bayesian model of label sequences with infinitely-long temporal dependencies, namely the sequence memoizer, and training our model using margin maximization and a versatile mean-field-like approximation to allow for increased computational efficiency. As we experimentally demonstrate, the devised margin-maximizing construction of our model, which leads to a convex optimization scheme, without any spurious local optima, combined with the capacity of our model to capture long and complex temporal dependencies, allow for obtaining exceptional pattern recognition performance in several applications. |
| Author | Chatzis, Sotirios P. |
| Author_xml | – sequence: 1 givenname: Sotirios P. surname: Chatzis fullname: Chatzis, Sotirios P. email: soteri0s@me.com organization: Department of Electrical Engineering, Computer Engineering and Informatics Cyprus University of Technology, Cyprus |
| BackLink | http://pascal-francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=27286037$$DView record in Pascal Francis |
| BookMark | eNqFkMFu1TAQRS1UJF4LP8AqGyQ2SW0nsWOJDaooVCpiA1usqTMu85TYD9ullK_H4VUsYWVpdO4dzzllJyEGZOyl4J3gQp3vO8z30Eku-o6Ljo_iCduJSfet0qY_YTtuRt0OQg_P2GnOe86F5lzv2NePkG4ptCv8pJV-Ubht3AI5kycHhWJoom8yfr_DUAiWZoYCzT2Vbw0FT4EKLg_tEmus4HqIaUPwgGHG4Ajzc_bUw5LxxeN7xr5cvvt88aG9_vT-6uLtdet6o0o7eINaO-Wkl8M0ocH5RhgpZw6gZzBO4KTMILGXCsSNGesY3WT0oKT3s-vP2Otj7yHF-tdc7ErZ4bJAwHiXbb1WbDeb8f9or0YxDkOvKyqPqEsx54TeHhKtkB6s4Hbzbvd2824375YLW73X0KvHfsgOFp-gmsh_k1LLSfE_5W-OHFYvPwiTzdVYcDhTQlfsHOlfa34Dvf6cfg |
| CitedBy_id | crossref_primary_10_1007_s00034_015_0220_4 crossref_primary_10_1016_j_engappai_2014_07_006 crossref_primary_10_1016_j_ins_2017_05_038 crossref_primary_10_3390_app12104841 crossref_primary_10_1016_j_neucom_2018_05_101 |
| Cites_doi | 10.1109/83.210863 10.1109/TPAMI.2005.221 10.1214/aop/1024404422 10.1109/TPAMI.2008.215 10.1007/0-387-28982-8 10.1109/ICASSP.1996.541126 10.1145/1897816.1897842 10.1214/aos/1176342360 10.1109/TFUZZ.2008.2005008 10.1109/CVPR.2007.383494 10.1016/S0031-3203(96)00106-9 10.1016/S0031-3203(02)00027-4 10.1109/TNN.2010.2046910 10.1016/j.patcog.2009.06.004 10.1109/ICASSP.1990.115992 10.1109/34.134040 10.1016/j.eswa.2012.02.193 10.1016/j.eswa.2009.02.050 10.5772/5816 10.1002/0471721182 10.1109/89.554265 10.1109/34.566806 10.1145/1553374.1553518 10.3115/1220175.1220299 10.1016/j.eswa.2009.06.063 10.1162/neco.1990.2.1.1 |
| ContentType | Journal Article |
| Copyright | 2013 Elsevier Ltd 2014 INIST-CNRS |
| Copyright_xml | – notice: 2013 Elsevier Ltd – notice: 2014 INIST-CNRS |
| DBID | IQODW AAYXX CITATION 7SC 8FD JQ2 L7M L~C L~D |
| DOI | 10.1016/j.eswa.2013.01.051 |
| DatabaseName | Pascal-Francis CrossRef Computer and Information Systems Abstracts Technology Research Database ProQuest Computer Science Collection Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Academic Computer and Information Systems Abstracts Professional |
| DatabaseTitle | CrossRef Computer and Information Systems Abstracts Technology Research Database Computer and Information Systems Abstracts – Academic Advanced Technologies Database with Aerospace ProQuest Computer Science Collection Computer and Information Systems Abstracts Professional |
| DatabaseTitleList | Computer and Information Systems Abstracts Computer and Information Systems Abstracts |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Computer Science Applied Sciences |
| EISSN | 1873-6793 |
| EndPage | 4527 |
| ExternalDocumentID | 10_1016_j_eswa_2013_01_051 27286037 S0957417413000821 |
| GroupedDBID | --K --M .DC .~1 0R~ 13V 1B1 1RT 1~. 1~5 4.4 457 4G. 5GY 5VS 7-5 71M 8P~ 9JN 9JO AAAKF AABNK AACTN AAEDT AAEDW AAIAV AAIKJ AAKOC AALRI AAOAW AAQFI AARIN AAXUO AAYFN ABBOA ABFNM ABMAC ABMVD ABUCO ABXDB ABYKQ ACDAQ ACGFS ACHRH ACNTT ACRLP ACZNC ADBBV ADEZE ADTZH AEBSH AECPX AEKER AENEX AFKWA AFTJW AGHFR AGJBL AGUBO AGUMN AGYEJ AHHHB AHJVU AHZHX AIALX AIEXJ AIKHN AITUG AJBFU AJOXV ALEQD ALMA_UNASSIGNED_HOLDINGS AMFUW AMRAJ AOUOD APLSM AXJTR BJAXD BKOJK BLXMC BNSAS CS3 DU5 EBS EFJIC EFLBG EJD EO8 EO9 EP2 EP3 F5P FDB FIRID FNPLU FYGXN G-Q GBLVA GBOLZ HAMUX IHE J1W JJJVA KOM LG9 LY1 LY7 M41 MO0 N9A O-L O9- OAUVE OZT P-8 P-9 P2P PC. PQQKQ Q38 RIG ROL RPZ SDF SDG SDP SDS SES SPC SPCBC SSB SSD SSL SST SSV SSZ T5K TN5 ~G- 08R 29G AAAKG AALMO AAPBV AAQXK ABKBG ABPIF ABPTK ACNNM ADALY ADJOM ASPBG AVWKF AZFZN FEDTE FGOYB G-2 HLZ HVGLF HZ~ IPNFZ IQODW PQEST R2- SBC SET SEW WUQ XPP ZMT AAXKI AAYXX ADMUD AFJKZ AKRWK CITATION 7SC 8FD JQ2 L7M L~C L~D |
| ID | FETCH-LOGICAL-c396t-4f9e77c6c2f2488e9edb1922d0aa7da9c1e86942e326a1b95aa7ec897462ffdc3 |
| ISSN | 0957-4174 |
| IngestDate | Fri Oct 25 02:55:20 EDT 2024 Sat Oct 26 00:15:11 EDT 2024 Thu Sep 26 16:49:34 EDT 2024 Fri Nov 25 13:52:51 EST 2022 Fri Feb 23 02:26:29 EST 2024 |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 11 |
| Keywords | Mean-field principle Sequential data modeling Margin maximization Sequence memoizer Costs Multiobjective programming Temporal databases Non parametric estimation Electronic countermeasure Modeling Convex programming Markov chain Mean-field theory Efficiency Vector support machine Convex function Learning algorithm Generative model Bayes estimation Local search Mean field approximation Data dependency Dependability Pattern recognition Trapping Data models Objective function Convex analysis |
| Language | English |
| License | CC BY 4.0 |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-c396t-4f9e77c6c2f2488e9edb1922d0aa7da9c1e86942e326a1b95aa7ec897462ffdc3 |
| Notes | ObjectType-Article-2 SourceType-Scholarly Journals-1 ObjectType-Feature-1 content type line 23 ObjectType-Article-1 ObjectType-Feature-2 |
| PQID | 1365154437 |
| PQPubID | 23500 |
| PageCount | 9 |
| ParticipantIDs | proquest_miscellaneous_1701001795 proquest_miscellaneous_1365154437 crossref_primary_10_1016_j_eswa_2013_01_051 pascalfrancis_primary_27286037 elsevier_sciencedirect_doi_10_1016_j_eswa_2013_01_051 |
| PublicationCentury | 2000 |
| PublicationDate | 2013-09-01 |
| PublicationDateYYYYMMDD | 2013-09-01 |
| PublicationDate_xml | – month: 09 year: 2013 text: 2013-09-01 day: 01 |
| PublicationDecade | 2010 |
| PublicationPlace | Amsterdam |
| PublicationPlace_xml | – name: Amsterdam |
| PublicationTitle | Expert systems with applications |
| PublicationYear | 2013 |
| Publisher | Elsevier Ltd Elsevier |
| Publisher_xml | – name: Elsevier Ltd – name: Elsevier |
| References | (pp. 2193–2196). Bishop (b0015) 2006 Celeux, Forbes, Peyrard (b0025) 2003; 36 Yuille (b0175) 1990; 2 Altun, Y., Tsochantaridis, I., & Hofmann, T. (2004). Hidden Markov support vector machines. In Jaakkola, Jordan (b0100) 1998 Chatzis, Demiris (b0035) 2012; 39 Zerubia, J., & Chellappa, R. (1990). Mean field approximation using compound Gauss–Markov random field for edge detection and image restoration. In Chatzis, Kosmopoulos, Varvarigou (b0040) 2009; 31 Teh, Y. W. (2006). A hierarchical Bayesian language model based on Pitman–Yor processes. In Gasthaus, J., & Teh, Y. W. (2011). Improvements to the sequence memoizer. In CMU. (2012). Hofmann, Buhmann (b0095) 1997; 19 New York. Vapnik (b0155) 1998 Wood, F., Archambeau, C., Gasthaus, J., James, L.F., & Teh, Y. (2009). A stochastic memoizer for sequence data. In Cappé, O., Moulines, E., & Rydén, T. (2005). Inference in hidden markov models. Zhang (b0190) 1993; 2 Aycard, Mari, Washington (b0010) 2004; 1 Eng, Thibessard, Hergalant, Mari, Leblond (b0075) 2005 (pp. 435–438). Sha, F., & Saul, L. K. (2006). Large margin Gaussian mixture modeling for phonetic classification and recognition. In Mari, J., Fohr, D., & Junqua, J. (1996). A second-order HMM for high-performance word and phoneme-based continuous speech recognition. In Pitman, Yor (b0135) 1997; 25 Dingand, Y., & Fan, G. (2007). Segmental hidden markov models for view-based sport video analysis. In Chatzis, Varvarigou (b0050) 2008; 16 Zeng, Duan, Wu (b0180) 2010 Nel, Preez, Herbst (b0130) 2005; 27 Zhao, Zhao, Zhu (b0195) 2009 Crammer, Singer (b0060) 2002; 2 Mari, Haton, Kriouile (b0120) 1997; 5 Srinivasan, Venkatesh, Hosie (b0145) 1997; 30 Engelbrecht, du Preez (b0070) 2010; 43 Ferguson (b0080) 1973; 1 . McLachlan, G., & Peel, D. (2000). Finite mixture models. Chandler (b0030) 1987 Ye, N., Lee, W.S., Chieu, H.L., Wu, D. & (2009). Conditional random fields with high-order features for sequence labeling. In Geiger, Girosi (b0090) 1991; 13 Lawrence, N. (2012). Gaussian process software. Wood, Gasthaus, Archambeau, James, Teh (b0165) 2011; 54 Chatzis, Tsechpenakis (b0045) 2010; 21 Luttrell (b0110) 1989 (pp. 265–268). Toulouse, France. (pp. 985–992). Chatzis (10.1016/j.eswa.2013.01.051_b0035) 2012; 39 Pitman (10.1016/j.eswa.2013.01.051_b0135) 1997; 25 Vapnik (10.1016/j.eswa.2013.01.051_b0155) 1998 10.1016/j.eswa.2013.01.051_b0105 Yuille (10.1016/j.eswa.2013.01.051_b0175) 1990; 2 Chandler (10.1016/j.eswa.2013.01.051_b0030) 1987 Nel (10.1016/j.eswa.2013.01.051_b0130) 2005; 27 10.1016/j.eswa.2013.01.051_b0055 Mari (10.1016/j.eswa.2013.01.051_b0120) 1997; 5 Luttrell (10.1016/j.eswa.2013.01.051_b0110) 1989 10.1016/j.eswa.2013.01.051_b0115 Zhang (10.1016/j.eswa.2013.01.051_b0190) 1993; 2 Bishop (10.1016/j.eswa.2013.01.051_b0015) 2006 Eng (10.1016/j.eswa.2013.01.051_b0075) 2005 Celeux (10.1016/j.eswa.2013.01.051_b0025) 2003; 36 Wood (10.1016/j.eswa.2013.01.051_b0165) 2011; 54 Hofmann (10.1016/j.eswa.2013.01.051_b0095) 1997; 19 10.1016/j.eswa.2013.01.051_b0150 10.1016/j.eswa.2013.01.051_b0170 Zeng (10.1016/j.eswa.2013.01.051_b0180) 2010 Crammer (10.1016/j.eswa.2013.01.051_b0060) 2002; 2 Jaakkola (10.1016/j.eswa.2013.01.051_b0100) 1998 Ferguson (10.1016/j.eswa.2013.01.051_b0080) 1973; 1 Aycard (10.1016/j.eswa.2013.01.051_b0010) 2004; 1 Zhao (10.1016/j.eswa.2013.01.051_b0195) 2009 Chatzis (10.1016/j.eswa.2013.01.051_b0045) 2010; 21 Chatzis (10.1016/j.eswa.2013.01.051_b0050) 2008; 16 Chatzis (10.1016/j.eswa.2013.01.051_b0040) 2009; 31 Geiger (10.1016/j.eswa.2013.01.051_b0090) 1991; 13 10.1016/j.eswa.2013.01.051_b0065 10.1016/j.eswa.2013.01.051_b0020 Srinivasan (10.1016/j.eswa.2013.01.051_b0145) 1997; 30 10.1016/j.eswa.2013.01.051_b0185 10.1016/j.eswa.2013.01.051_b0005 10.1016/j.eswa.2013.01.051_b0125 Engelbrecht (10.1016/j.eswa.2013.01.051_b0070) 2010; 43 10.1016/j.eswa.2013.01.051_b0085 10.1016/j.eswa.2013.01.051_b0140 10.1016/j.eswa.2013.01.051_b0160 |
| References_xml | – volume: 21 start-page: 1004 year: 2010 end-page: 1014 ident: b0045 article-title: The infinite hidden Markov random field model publication-title: IEEE Transactions on Neural Networks contributor: fullname: Tsechpenakis – start-page: 1550 year: 2010 end-page: 1555 ident: b0180 article-title: A new distance measure for hidden Markov models publication-title: Expert Systems with Applications contributor: fullname: Wu – volume: 2 start-page: 265 year: 2002 end-page: 292 ident: b0060 article-title: On the algorithmic implementation of multi-class SVMs publication-title: Journal of Machine Learning Research contributor: fullname: Singer – volume: 43 start-page: 99 year: 2010 end-page: 112 ident: b0070 article-title: Efficient backward decoding of high-order hidden markov models publication-title: Pattern Recognition contributor: fullname: du Preez – volume: 1 start-page: 231 year: 2004 end-page: 245 ident: b0010 article-title: Learning to automatically detect features for mobile robots using second-order hidden Markov models publication-title: International Journal of Advanced Robotic System contributor: fullname: Washington – volume: 1 start-page: 209 year: 1973 end-page: 230 ident: b0080 article-title: A Bayesian analysis of some nonparametric problems publication-title: The Annals of Statistics contributor: fullname: Ferguson – volume: 30 start-page: 593 year: 1997 end-page: 606 ident: b0145 article-title: Qualitative estimation of camera motion parameters from video sequences publication-title: Pattern Recognition contributor: fullname: Hosie – volume: 39 start-page: 10303 year: 2012 end-page: 10309 ident: b0035 article-title: The echo state conditional random field model for sequential data modeling publication-title: Expert Systems with Applications contributor: fullname: Demiris – volume: 25 start-page: 855 year: 1997 end-page: 900 ident: b0135 article-title: The two-parameter Poisson–Dirichlet distribution derived from a stable subordinator publication-title: Annals of Probability contributor: fullname: Yor – volume: 13 start-page: 401 year: 1991 end-page: 412 ident: b0090 article-title: Parallel and deterministic algorithms from MRFs: surface reconstruction publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence contributor: fullname: Girosi – volume: 36 start-page: 131 year: 2003 end-page: 144 ident: b0025 article-title: EM procedures using mean field-like approximations for Markov model-based image segmentation publication-title: Pattern Recognition contributor: fullname: Peyrard – volume: 16 start-page: 1351 year: 2008 end-page: 1361 ident: b0050 article-title: A fuzzy clustering approach toward hidden Markov random field models for enhanced spatially constrained image segmentation publication-title: IEEE Transactions on Fuzzy Systems contributor: fullname: Varvarigou – start-page: 163 year: 1998 end-page: 173 ident: b0100 article-title: Improving the mean field approximation via the use of mixture distributions publication-title: Learning in graphical models contributor: fullname: Jordan – year: 2005 ident: b0075 article-title: Data mining using hidden Markov models (HMM2) to detect heterogeneities into bacteria genomes publication-title: Journes Ouvertes Biologie Informatique Mathematiques (JOBIM) contributor: fullname: Leblond – volume: 2 start-page: 1 year: 1990 end-page: 24 ident: b0175 article-title: Generalized deformable models, statistical physics and matching problems publication-title: Neural Computation contributor: fullname: Yuille – year: 1998 ident: b0155 article-title: Statistical learning theory contributor: fullname: Vapnik – volume: 27 start-page: 1733 year: 2005 end-page: 1746 ident: b0130 article-title: Estimating the pen trajectories of static signatures using hidden Markov models publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence contributor: fullname: Herbst – volume: 2 start-page: 27 year: 1993 end-page: 40 ident: b0190 article-title: The mean field theory in EM procedures for Markov random fields publication-title: IEEE Transactions on Image Processing contributor: fullname: Zhang – year: 1987 ident: b0030 article-title: Introduction to modern statistical mechanics contributor: fullname: Chandler – start-page: 363 year: 1989 end-page: 370 ident: b0110 article-title: The use of Bayesian and entropic methods in neural network theory publication-title: Maximum entropy and bayesian methods contributor: fullname: Luttrell – year: 2006 ident: b0015 article-title: Pattern recognition and machine learning contributor: fullname: Bishop – volume: 5 start-page: 22 year: 1997 end-page: 25 ident: b0120 article-title: Automatic word recognition based on second-order hidden Markov models publication-title: IEEE Transanctions on Speech and Audio Processing contributor: fullname: Kriouile – volume: 54 start-page: 91 year: 2011 end-page: 98 ident: b0165 article-title: The sequence memoizer publication-title: Communications of the ACM contributor: fullname: Teh – start-page: 9813 year: 2009 end-page: 9818 ident: b0195 article-title: TSVM-HMM: Transductive SVM based hidden Markov model for automatic image annotation publication-title: Expert Systems with Applications contributor: fullname: Zhu – volume: 31 start-page: 1657 year: 2009 end-page: 1669 ident: b0040 article-title: Robust sequential data modeling using an outlier tolerant hidden Markov model publication-title: IEEE Transactions on Pattern Analysis and Machine Intelligence contributor: fullname: Varvarigou – volume: 19 start-page: 1 year: 1997 end-page: 14 ident: b0095 article-title: Pairwise data clustering by deterministic annealing publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence contributor: fullname: Buhmann – start-page: 363 year: 1989 ident: 10.1016/j.eswa.2013.01.051_b0110 article-title: The use of Bayesian and entropic methods in neural network theory contributor: fullname: Luttrell – ident: 10.1016/j.eswa.2013.01.051_b0170 – volume: 2 start-page: 27 issue: 1 year: 1993 ident: 10.1016/j.eswa.2013.01.051_b0190 article-title: The mean field theory in EM procedures for Markov random fields publication-title: IEEE Transactions on Image Processing doi: 10.1109/83.210863 contributor: fullname: Zhang – volume: 27 start-page: 1733 issue: 11 year: 2005 ident: 10.1016/j.eswa.2013.01.051_b0130 article-title: Estimating the pen trajectories of static signatures using hidden Markov models publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence doi: 10.1109/TPAMI.2005.221 contributor: fullname: Nel – volume: 25 start-page: 855 year: 1997 ident: 10.1016/j.eswa.2013.01.051_b0135 article-title: The two-parameter Poisson–Dirichlet distribution derived from a stable subordinator publication-title: Annals of Probability doi: 10.1214/aop/1024404422 contributor: fullname: Pitman – ident: 10.1016/j.eswa.2013.01.051_b0005 – volume: 31 start-page: 1657 issue: 9 year: 2009 ident: 10.1016/j.eswa.2013.01.051_b0040 article-title: Robust sequential data modeling using an outlier tolerant hidden Markov model publication-title: IEEE Transactions on Pattern Analysis and Machine Intelligence doi: 10.1109/TPAMI.2008.215 contributor: fullname: Chatzis – volume: 2 start-page: 265 year: 2002 ident: 10.1016/j.eswa.2013.01.051_b0060 article-title: On the algorithmic implementation of multi-class SVMs publication-title: Journal of Machine Learning Research contributor: fullname: Crammer – ident: 10.1016/j.eswa.2013.01.051_b0020 doi: 10.1007/0-387-28982-8 – ident: 10.1016/j.eswa.2013.01.051_b0115 doi: 10.1109/ICASSP.1996.541126 – start-page: 163 year: 1998 ident: 10.1016/j.eswa.2013.01.051_b0100 article-title: Improving the mean field approximation via the use of mixture distributions contributor: fullname: Jaakkola – ident: 10.1016/j.eswa.2013.01.051_b0105 – volume: 54 start-page: 91 issue: 2 year: 2011 ident: 10.1016/j.eswa.2013.01.051_b0165 article-title: The sequence memoizer publication-title: Communications of the ACM doi: 10.1145/1897816.1897842 contributor: fullname: Wood – volume: 1 start-page: 209 year: 1973 ident: 10.1016/j.eswa.2013.01.051_b0080 article-title: A Bayesian analysis of some nonparametric problems publication-title: The Annals of Statistics doi: 10.1214/aos/1176342360 contributor: fullname: Ferguson – volume: 16 start-page: 1351 issue: 5 year: 2008 ident: 10.1016/j.eswa.2013.01.051_b0050 article-title: A fuzzy clustering approach toward hidden Markov random field models for enhanced spatially constrained image segmentation publication-title: IEEE Transactions on Fuzzy Systems doi: 10.1109/TFUZZ.2008.2005008 contributor: fullname: Chatzis – ident: 10.1016/j.eswa.2013.01.051_b0055 – ident: 10.1016/j.eswa.2013.01.051_b0065 doi: 10.1109/CVPR.2007.383494 – volume: 30 start-page: 593 year: 1997 ident: 10.1016/j.eswa.2013.01.051_b0145 article-title: Qualitative estimation of camera motion parameters from video sequences publication-title: Pattern Recognition doi: 10.1016/S0031-3203(96)00106-9 contributor: fullname: Srinivasan – year: 2006 ident: 10.1016/j.eswa.2013.01.051_b0015 contributor: fullname: Bishop – year: 2005 ident: 10.1016/j.eswa.2013.01.051_b0075 article-title: Data mining using hidden Markov models (HMM2) to detect heterogeneities into bacteria genomes publication-title: Journes Ouvertes Biologie Informatique Mathematiques (JOBIM) contributor: fullname: Eng – volume: 36 start-page: 131 issue: 1 year: 2003 ident: 10.1016/j.eswa.2013.01.051_b0025 article-title: EM procedures using mean field-like approximations for Markov model-based image segmentation publication-title: Pattern Recognition doi: 10.1016/S0031-3203(02)00027-4 contributor: fullname: Celeux – volume: 21 start-page: 1004 issue: 6 year: 2010 ident: 10.1016/j.eswa.2013.01.051_b0045 article-title: The infinite hidden Markov random field model publication-title: IEEE Transactions on Neural Networks doi: 10.1109/TNN.2010.2046910 contributor: fullname: Chatzis – volume: 43 start-page: 99 issue: 1 year: 2010 ident: 10.1016/j.eswa.2013.01.051_b0070 article-title: Efficient backward decoding of high-order hidden markov models publication-title: Pattern Recognition doi: 10.1016/j.patcog.2009.06.004 contributor: fullname: Engelbrecht – ident: 10.1016/j.eswa.2013.01.051_b0185 doi: 10.1109/ICASSP.1990.115992 – volume: 13 start-page: 401 issue: 5 year: 1991 ident: 10.1016/j.eswa.2013.01.051_b0090 article-title: Parallel and deterministic algorithms from MRFs: surface reconstruction publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence doi: 10.1109/34.134040 contributor: fullname: Geiger – volume: 39 start-page: 10303 issue: 11 year: 2012 ident: 10.1016/j.eswa.2013.01.051_b0035 article-title: The echo state conditional random field model for sequential data modeling publication-title: Expert Systems with Applications doi: 10.1016/j.eswa.2012.02.193 contributor: fullname: Chatzis – start-page: 9813 year: 2009 ident: 10.1016/j.eswa.2013.01.051_b0195 article-title: TSVM-HMM: Transductive SVM based hidden Markov model for automatic image annotation publication-title: Expert Systems with Applications doi: 10.1016/j.eswa.2009.02.050 contributor: fullname: Zhao – ident: 10.1016/j.eswa.2013.01.051_b0085 – volume: 1 start-page: 231 issue: 4 year: 2004 ident: 10.1016/j.eswa.2013.01.051_b0010 article-title: Learning to automatically detect features for mobile robots using second-order hidden Markov models publication-title: International Journal of Advanced Robotic System doi: 10.5772/5816 contributor: fullname: Aycard – ident: 10.1016/j.eswa.2013.01.051_b0125 doi: 10.1002/0471721182 – year: 1998 ident: 10.1016/j.eswa.2013.01.051_b0155 contributor: fullname: Vapnik – volume: 5 start-page: 22 issue: 3 year: 1997 ident: 10.1016/j.eswa.2013.01.051_b0120 article-title: Automatic word recognition based on second-order hidden Markov models publication-title: IEEE Transanctions on Speech and Audio Processing doi: 10.1109/89.554265 contributor: fullname: Mari – volume: 19 start-page: 1 issue: 1 year: 1997 ident: 10.1016/j.eswa.2013.01.051_b0095 article-title: Pairwise data clustering by deterministic annealing publication-title: IEEE Transanctions on Pattern Analysis and Machine Intelligence doi: 10.1109/34.566806 contributor: fullname: Hofmann – ident: 10.1016/j.eswa.2013.01.051_b0160 doi: 10.1145/1553374.1553518 – ident: 10.1016/j.eswa.2013.01.051_b0150 doi: 10.3115/1220175.1220299 – start-page: 1550 year: 2010 ident: 10.1016/j.eswa.2013.01.051_b0180 article-title: A new distance measure for hidden Markov models publication-title: Expert Systems with Applications doi: 10.1016/j.eswa.2009.06.063 contributor: fullname: Zeng – ident: 10.1016/j.eswa.2013.01.051_b0140 – year: 1987 ident: 10.1016/j.eswa.2013.01.051_b0030 contributor: fullname: Chandler – volume: 2 start-page: 1 year: 1990 ident: 10.1016/j.eswa.2013.01.051_b0175 article-title: Generalized deformable models, statistical physics and matching problems publication-title: Neural Computation doi: 10.1162/neco.1990.2.1.1 contributor: fullname: Yuille |
| SSID | ssj0017007 |
| Score | 2.1721199 |
| Snippet | ► We present a method for sequential data modeling. ► Our approach models temporal dependencies of infinite length. ► It employs a margin maximization training... Generative models for sequential data are usually based on the assumption of temporal dependencies described by a first-order Markov chain. To ameliorate this... |
| SourceID | proquest crossref pascalfrancis elsevier |
| SourceType | Aggregation Database Index Database Publisher |
| StartPage | 4519 |
| SubjectTerms | Algorithms Applied sciences Computational efficiency Computer science; control theory; systems Data processing. List processing. Character string processing Decision theory. Utility theory Exact sciences and technology Information systems. Data bases Margin maximization Mathematical analysis Mathematical models Mathematical programming Maximization Mean-field principle Memory organisation. Data processing Military technology Operational research and scientific management Operational research. Management science Sequence memoizer Sequential data modeling Software Temporal logic Training |
| Title | Margin-maximizing classification of sequential data with infinitely-long temporal dependencies |
| URI | https://dx.doi.org/10.1016/j.eswa.2013.01.051 https://search.proquest.com/docview/1365154437 https://search.proquest.com/docview/1701001795 |
| Volume | 40 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://sdu.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV3di9QwEA969yKI3-L6cUTwbcnS7zSPh66oDyLsCfdkyCYpdNlrj20X9f56ZzZN273FQwVfSgm7TZr5dTKZzPyGkDewIocmzBXTSbBkCayBLAfDlRUiS5ZGqMSYnetiwT-f5-_myXyIVR3a_qukoQ1kjZmzfyHt_qHQAPcgc7iC1OH6R3LHyrVlxS7Uj_KivNql1KKBjBFBvXXo4qdbdJZjhKhzxkLnJRqg659sjQWIOtKq9dTXydU-3nDVh-_ZTdtxQfssudF5-ChyoL1yVAaLui03Zd1Mv8zG7gYs_SDG7oY-D2YIOnLORM6S0NXbmVmnSnMes4y7-ode1zpqJo-pcKQ5keZmtAonqaMMONDwztmwmtnmO9JGhfGOdbUjrd1nzl7gsHBUeGQHpg5sko8jGBCow-PTj_PzT_1xEw9cXr1_jS67ygUCXu_pdxbM3UvVwHdVuIIoB2v7zmA5e0DudTsNeuog8pDcstUjct9X8aCdUn9Mvh0ghu4jhtYFHRBDETEUhU2vIYZ6xNAxYp6Qr-_nZ28_sK7qBtOxyFqWFMJyrjMdFRFodyusWcI2IDKBUtwooUObZyKJLBj-KlyKFJqtzmFfmkVFYXT8lBxVdWWfEWpFEC_zVPMiEkmWF8rGNs51yo0tQh3qCZn6mZSXjlxF-qjDlcR5lzjvMgglzPuEpH6yZWceOrNPAjZu_N_JnmT6riIe5VkQ8wl57UUlQbniiZmqbL1tJMaAIl3Vjb_hAfKYcZE-_8cBviB3hi_tJTlqN1v7itxuzPakA-ovv2avjg |
| link.rule.ids | 315,782,786,27933,27934 |
| linkProvider | Elsevier |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Margin-maximizing+classification+of+sequential+data+with+infinitely-long+temporal+dependencies&rft.jtitle=Expert+systems+with+applications&rft.au=Chatzis%2C+Sotirios+P.&rft.date=2013-09-01&rft.pub=Elsevier+Ltd&rft.issn=0957-4174&rft.eissn=1873-6793&rft.volume=40&rft.issue=11&rft.spage=4519&rft.epage=4527&rft_id=info:doi/10.1016%2Fj.eswa.2013.01.051&rft.externalDocID=S0957417413000821 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0957-4174&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0957-4174&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0957-4174&client=summon |