Morphological segmentation using dirichlet process based bayesian non-parametric models

Download

index.pdf

Date

2016

Author

Kumyol, Serkan

Metadata

Show full item record

Item Usage Stats

200
views

83
downloads

This study, will try to explore models explaining distributional properties of morphology within the morphological segmentation task. There are di erent learning approaches to the morphological segmentation task based on supervised, semi-supervised and unsupervised learning. The existing systems and how well semi-supervised and unsupervised non-parametric Bayesian models t to the segmentation task will be investigated. Furthermore, the role of occurrence independent and co-occurrence based models in morpheme segmentation will be investigated.

Subject Keywords

Computational linguistics., Natural language processing (Computer science)., Grammar, Comparative and general

URI

http://etd.lib.metu.edu.tr/upload/12619892/index.pdf
https://hdl.handle.net/11511/25564

Collections

Graduate School of Informatics, Thesis

Suggestions

OpenMETU
Core

Structural priming in Turkish genitive-possessive constructions Bahadır, Gözde; Hohenberger, Annette Edeltraud; Zeyrek Bozşahin, Deniz; Department of Cognitive Sciences (2012) This study addresses the question of the mental representation and processing of language by investigating “structural priming” in Turkish Genitive-Possessive (GEN-POSS) constructions. Structural priming is the facilitating effect of having already experienced a structural form on its subsequent processing. We investigate this phenomenon on a construction pair in Turkish, which shares the same external GEN-POSS morpho-syntactic template despite having distinct grammatical categories. The structures under sc...
Multi-subject brain decoding using deep learning techniques Velioğlu, Burak; Yarman Vural, Fatoş Tunay; Ertekin Bolelli, Şeyda; Department of Computer Engineering (2016) In this study, a new method is proposed for analyzing and classifying images obtained by functional magnetic resonance imaging (fMRI) from multiple subjects. Considering the multi level structure of the brain and success of deep learning architectures on extracting hierarchical representations from raw data, these architectures are used in this thesis. Initially, the S500 data set collected in the scope of Human Connectome Project (HCP) is used to train formed deep neural networks in an unsupervised fashion...
An Experimental study on acquisition of prepositions in English as a third language Çabuk, Sakine; Sağın Şimşek, Sultan Çiğdem; Gracanın Yüksek, Martına; Department of English Language Teaching (2016) This study explores the role of cross-linguistic influence in third language acquisition process by examining English adpositions. Comprehension, processing and production of English prepositions (in, on, at, behind, over, to) were examined through off-line and on-line data collection tasks to find out which of the two known languages (L1 or L2) is the major source of cross-linguistic influence on the acquisition of English (L3) adpositions given the fact that adpositions are morphologically and syntactical...
Localization and identification of structural nonlinearities using cascaded optimization and neural networks Koyuncu, A.; Ciğeroğlu, Ender; Özgüven, Hasan Nevzat (Elsevier BV, 2017-10-01) In this study, a new approach is proposed for identification of structural nonlinearities by employing cascaded optimization and neural networks. Linear finite element model of the system and frequency response functions measured at arbitrary locations of the system are used in this approach. Using the finite element model, a training data set is created, which appropriately spans the possible nonlinear configurations space of the system. A classification neural network trained on these data sets then local...
Performance of a non-linear adaptive beamformer algorithm for signal-of-interest extraction / Oğuz, Özkan; Tuncer, Temel Engin; Department of Electrical and Electronics Engineering (2015) In this thesis a non-linear adaptive beamforming technique, Adaptive Projections Subgradient Method [1] (APSM) is considered. This method uses projections over convex sets in Reproducing Kernel Hilbert Space. Main advantage of this method is observed if the signal-of-interest is due to digital modulation and when there are more jammers than the number of antennas. The performance of this non-linear beamforming technique is compared with well-known methods including Minimum Variance Distortionless Response [...

Citation Formats

S. Kumyol, “Morphological segmentation using dirichlet process based bayesian non-parametric models,” M.S. - Master of Science, Middle East Technical University, 2016.