Research interests : language learning, NLP word embeddings and semantic representations in languages, computer vision.
Work
Sem2Vec: Semantic Word Vector Specializations with Bidirectional Constraint Propagations
[PDF]
@ARTICLE{Kekec2019_TKDE,
author={T. {Kekeç} and D. M. {Tax}},
journal={IEEE Transactions on Knowledge and Data Engineering},
title={Sem2Vec: Semantic Word Vectors with Bidirectional Constraint Propagations},
year={2019},
volume={},
number={},
pages={1-1},
keywords={Semantics;Dictionaries;Task analysis;Thesauri;Dogs;Buildings;Stability analysis;Word Embeddings;Semantic Embeddings;Embedding Stability;Thesaurus;Constraint Propagation},
doi={10.1109/TKDE.2019.2942021},
ISSN={},
month={}, }
Taygun Kekeç, D. M. J. Tax IEEE-TKDE: IEEE Transactions on Knowledge and Data Engineering , 2019.
Boosted Negative Sampling by Quadratically Constrained Entropy Maximization
[PDF]
[SUPP]
@article{Kekec2019_PRL,
title = "Boosted negative sampling by quadratically constrained entropy maximization",
journal = "Pattern Recognition Letters",
volume = "125",
pages = "310 - 317",
year = "2019",
issn = "0167-8655",
doi = "https://doi.org/10.1016/j.patrec.2019.04.027",
url = "http://www.sciencedirect.com/science/article/pii/S0167865519301424",
author = "Taygun Kekeç and David Mimno and David M.J. Tax",
keywords = "Word embeddings, Contrastive learning, Negative sampling, Entropy maximization, Semantic similarity",
abstract = "Learning probability densities for natural language representations is a difficult problem because language is inherently sparse and high-dimensional. Negative sampling is a popular and effective way to avoid intractable maximum likelihood problems, but it requires correct specification of the sampling distribution. Previous state of the art methods rely on heuristic distributions that appear to do well in practice. In this work, we define conditions for optimal sampling distributions and demonstrate how to approximate them using Quadratically Constrained Entropy Maximization(QCEM). Our analysis shows that state of the art heuristics are restrictive approximations to our proposed framework. To demonstrate the merits of our formulation, we apply QCEM to matching synthetic exponential family distributions and to finding high dimensional word embedding vectors for English. We are able to achieve faster inference on synthetic experiments and improve the correlation on semantic similarity evaluations on the Rare Words dataset by 4.8%."
}
Taygun Kekeç, David Mimno, D. M. J. Tax PRL: Journal of Pattern Recognition Letters , 2019.
@inproceedings{Kekec2018_ICISDM,
author = {Keke\c{c}, Taygun and van der Maaten, Laurens and Tax, D. M. J.},
title = {PAWE: Polysemy Aware Word Embeddings},
booktitle = {Proceedings of the 2Nd International Conference on Information System and Data Mining},
series = {ICISDM '18},
year = {2018},
isbn = {978-1-4503-6354-9},
location = {Lakeland, FL, USA},
pages = {7--13},
numpages = {7},
url = {http://doi.acm.org/10.1145/3206098.3206101},
doi = {10.1145/3206098.3206101},
acmid = {3206101},
publisher = {ACM},
address = {New York, NY, USA},
keywords = {Contrastive Estimation, Loglinear Models, Polysemy, Semantic Similarity, Word Embeddings},
}
Taygun Kekeç, Laurens van der Maaten, D. M. J. Tax ICISDM: International Conference on Information Systems and Data Mining , 2018.
@INPROCEEDINGS{Çetinay2017_SEGE,
author={H. Cetinay and T. Kekeç and F. A. Kuipers and D. M. J. Tax},
booktitle={2017 IEEE International Conference on Smart Energy Grid Engineering (SEGE)},
title={Markov Random Field for wind farm planning},
year={2017},
volume={},
number={},
pages={182-187},
keywords={Markov processes;decision making;fuzzy set theory;investment;power generation planning;wind power plants;Markov Random Field;coarse wind farm suitability value;fuzzy sets;multiple-criteria decision analysis;wind energy;wind farm investment;wind farm planning;wind profiles;Decision analysis;Investment;Markov random fields;Planning;Wind farms;Wind power generation;criteria for wind farms;planning for wind power integration;quality of wind;spatial relations in wind farm investments},
doi={10.1109/SEGE.2017.8052796},
ISSN={},
month={Aug}
}
Hâle Çetinay, Taygun Kekeç, Fernando A. Kuipers, D. M. J. Tax SEGE: International Conference on Smart Energy Grid Engineering, 2017.
@inproceedings{DBLP:conf/emnlp/KekecT16,
author = {Taygun Kekeç and
David M. J. Tax},
title = {Robust Gram Embeddings},
booktitle = {Proceedings of the 2016 Conference on Empirical Methods in Natural
Language Processing, {EMNLP} 2016, Austin, Texas, USA, November 1-4,
2016},
pages = {1060--1065},
year = {2016},
crossref = {DBLP:conf/emnlp/2016},
url = {http://aclweb.org/anthology/D/D16/D16-1113.pdf},
timestamp = {Fri, 04 Nov 2016 14:45:31 +0100},
biburl = {http://dblp.uni-trier.de/rec/bib/conf/emnlp/KekeçT16},
bibsource = {dblp computer science bibliography, http://dblp.org}
}
Taygun Kekeç, D. M. J. Tax EMNLP: Empirical Methods in Natural Language Processing, 2016.
Contextually Constrained Deep Networks for Scene Labeling
[PDF]
@inproceedings{Kekec2014contextually,
title={Contextually Constrained Deep Networks for Scene Labeling},
author={Kekeç, Taygun and Emonet, R{\'e}mi and Fromont, Elisa and Tr{\'e}meau, Alain and Wolf, Christian and Saint-Etienne, France},
booktitle={Proceedings of the British Machine Vision Conference, 2014},
year={2014}
}
Taygun Kekeç, Rémi Emonet, Elisa Fromont, Alain Trémeau, Christian Wolf BMVC: British Machine Vision Conference, 2014.
Prise en Compte du Contexte pour Contraindre les Réseaux Profonds
[PDF]
Taygun Kekeç, Rémi Emonet, Elisa Fromont, Alain Trémeau, Christian Wolf CAP: Conférence d'Apprentissage Automatique, 2014.
A New Approach to Real-time Mosaicing of Aerial Images
[Video]
[PDF]
@article{Kekec20141755,
title = "A new approach to real-time mosaicing of aerial images ",
journal = "Robotics and Autonomous Systems ",
volume = "62",
number = "12",
pages = "1755 - 1767",
year = "2014",
note = "",
issn = "0921-8890",
doi = "http://dx.doi.org/10.1016/j.robot.2014.07.010",
url = "http://www.sciencedirect.com/science/article/pii/S0921889014001377",
author = "Taygun Kekeç and Alper Yıldırım and Mustafa Ünel",
keywords = "Image mosaicing",
keywords = "Bundle adjustment",
keywords = "\{MLESAC\}",
keywords = "Separating axis theorem",
keywords = "Affine refinement",
keywords = "Multi-band blending",
keywords = "Gain compensation "
Taygun Kekeç, Alper Yıldırım, Mustafa Ünel Robotics and Autonomous Systems, 2014.
A Modular Software Architecture for UAVs
[PDF]
[Best]
@Article{Taygun2013_IECON,
Author = {T. Kekeç and B. C. Üstündağ and M. A. Güney and A. Yıldırım and M. Ünel},
Title = {A Modular Software Architecture for UAVs},
BookTitle = {39th {A}nnual {C}onference of the {IEEE} {I}ndustrial {E}lectronics {S}ociety},
year = {2013},
pages = {4037-4042}
}
Taygun Kekeç, Barış Can Üstündağ, Mehmet Ali Güney, Alper Yıldırım, Mustafa Ünel. IECON: 39th Annual Conference of the IEEE Industrial Electronics Society, 2013.
Real-time Multiple Object Tracking using Virtual Shells
@Article{Taygun2013_ICMCA,
Author = {T. Kekeç and and M. Ünel and H. Erdoğan},
Title = {Real-time Multiple Object Tracking using Virtual Shells},
BookTitle = {{I}nternational {C}onference on {M}athemathical and {C}omputational {A}pplications},
year = {2013},
volume = {18},
number = {3},
pages = {373}
}
Taygun Kekeç, Mustafa Ünel, Hakan Erdoğan ICMCA: International Conference on Mathematical and Computational Applications, 2013.
İnsansız Hava Araçları İçin Donanımdan Bağımsız Yazılım Geliştirilmesi
Barış Can Üstündağ, Taygun Kekeç, Mehmet Ali Güney, Pamir Mundt, Mustafa Ünel TOK: Otomatik Kontrol Ulusal Toplantısı, 2013.
İnsansız Hava Araçları İçin Modüler Bir Sistem Mimarisi
Taygun Kekeç, Barış Can Üstündağ, Mustafa Ünel TOK: Otomatik Kontrol Ulusal Toplantısı, 2012.