SciTePress - Affiliation Profile

Kocaeli University, Turkey

Context (please use complete words only)

Field

Sort by

6 Papers

An Investigation of Multi-Language Age Classification from Voice

Topics: Neural Networks for Biosignal Data; Pattern Recognition & Machine Learning for Biosignal Data

Authors: Osman Büyük ¹ and Levent M. Arslan ^{2

;

3}

Affiliations: ¹ Department of Electronics and Communications Engineering, Kocaeli University, Kocaeli, Turkey , ² Department of Electrical and Electronics Engineering, Bogazici University, Istanbul, Turkey , ³ Sestek Speech Enabled Software Technologies Incorporation, Istanbul, Turkey

Digital Object Identifier: 10.5220/0007237600850092

Proceedings of BIOSIGNALS 2019/BIOSTEC, ISBN: 978-989-758-353-7

Quick Abstract

In this paper, we investigate the use of deep neural networks (DNN) for a multi-language age classification task using speaker’s voice. For this purpose, speech databases in two different languages are combined together to construct a multi-language database. Mel-frequency cepstral coefficients (MFCC) are extracted for each utterance. A Gaussian mixture model (GMM), a support vector machine (SVM) and a feed-forward deep neural network (DNN) systems are trained using the features. In the SVM and DNN methods, the GMM means are concatenated to obtain a GMM supervector. The supervectors are fed into the SVM and DNN for age classification. In the experiments, we observe that the multi-language training does not degrade the performance in the SVM and DNN methods when compared to the matched training where train and test languages are the same. On the other hand, the performance is degraded for the traditional GMM method. Additionally, the SVM and DNN significantly outperform the GMM in the multi-language train-test scenario. The absolute performance improvement with the SVM and DNN is approximately 12% and 7% for female and male speakers, respectively.

Extracting Implicit Aspects based on Latent Dirichlet Allocation

Authors: Ekin Ekinci and Sevinç İlhan Omurca

Affiliation: Kocaeli University, Turkey

Proceedings of DCAART 2017/ICAART

Quick Abstract

Sentiment analysis arises as one of the important research field. With the growing popularity of opinion-rich resources such as ecommerce and social media websites, blogs, dictionaries and news portals people and companies start to understand the opinions of others by using this mediums. The majority of the studies on sentiment analysis focus on whether or not the meaning of the text is positive or negative. Nowadays, aspect based sentiment analysis has become a prominent field of study for in-depth analysis of customer reviews. In sentiment analysis aspects are categorized as two types: explicit aspects and implicit aspects. In this study we aim to extract implicit aspects with topic models.

Combining Spectral and Prosodic Features in HMM-based Single Utterance Speaker Verification

Topics: Speech Recognition; Neural Networks; Biometrics; Pattern Recognition; Acoustic Signal Processing

Authors: Osman Büyük ¹ and Levent M. Arslan ²

Affiliations: ¹ Kocaeli University, Turkey , ² Bogazici University and Sestek Inc., Turkey

Digital Object Identifier: 10.5220/0005217500860091

Proceedings of BIOSIGNALS 2015/BIOSTEC, ISBN: 978-989-758-069-7

Quick Abstract

In this paper, we combine spectral and prosodic features together in order to improve the verification performance on a text-dependent single utterance speaker verification task. The baseline spectral system makes use of a whole-phrase sentence HMM topology for the fixed utterance. We extract prosodic features using time alignment information obtained from the HMM states. In our experiments we observe that, although the prosodic features individually do not yield high performance, they provide complementary information to the spectral features. We achieve approximately 10% relative reduction in EER when the information sources are combined with a multi-layer neural network.

Related Ontology Subjects/Areas/Topics (25)

Acoustic Signal Processing	Health Engineering and Technology Applications	Pattern Recognition
Applications	Human-Computer Interaction	Physiological Computing Systems
Artificial Intelligence	Methodologies and Methods	Sensor Networks
Biomedical Engineering	Multimedia	Signal Processing
Biomedical Signal Processing	Multimedia Signal Processing	Soft Computing
Biometrics	Neural Networks	Speech Recognition
Biometrics and Pattern Recognition	Neurocomputing	Telecommunications
Computational Intelligence	Neurotechnology, Electronics and Informatics	Theory and Methods
Data Manipulation

A Performance Evaluation Model of a Job Title using Fuzzy Approach

Topics: Complex Fuzzy Systems; Industrial, Financial and Medical Applications

Authors: Hatice Esen , Tuğçen Hatipoğlu and Ali İhsan Boyacı

Affiliation: Kocaeli University, Turkey

Digital Object Identifier: 10.5220/0005076500530060

Proceedings of FCTA 2014/IJCCI, ISBN: 978-989-758-053-6

Quick Abstract

Performance evaluation is described as comparing the performance of workers and the work standards and handling the necessary activities in a systematic way to attain these standards. What makes performance measure a necessity is its focus on performance of personnel as an objective measure of whether the company goes in the correct direction or not. This is because the most important problem encountered in organizations is the difficulty in the determination of how successful the personnel are in the satisfaction of their duties and what are their capabilities in their jobs. Besides performance evaluation is a decision making process which involves uncertainty. To overcome the uncertainty and evaluate the workers performance objectively, a performance evaluation model is developed of which the criteria are defined as the fuzzy numbers and the linguistic variables. The scope of the study is to determine the performance evaluation criteria of a purchasing specialist and weight for evaluating this job title.

Related Ontology Subjects/Areas/Topics (21)

Artificial Intelligence	Fuzzy Systems	Neurotechnology, Electronics and Informatics
Biomedical Engineering	Health Engineering and Technology Applications	Pattern Recognition
Biomedical Signal Processing	Human-Computer Interaction	Physiological Computing Systems
Complex Fuzzy Systems	Industrial, Financial and Medical Applications	Sensor Networks
Computational Intelligence	Methodologies and Methods	Signal Processing
Computer-Supported Education	Neural Networks	Soft Computing
Domain Applications and Case Studies	Neurocomputing	Theory and Methods

Supplier Selection Using Fuzzy Linear Programming Technique for Multidimensional Analysis of Preference (LINMAP)

Topics: Complex Fuzzy Systems; Industrial, Financial and Medical Applications

Authors: Ali İhsan Boyacı , Tuğçen Hatipoğlu and Hatice Esen

Affiliation: Kocaeli University, Turkey

Digital Object Identifier: 10.5220/0005076601190126

Proceedings of FCTA 2014/IJCCI, ISBN: 978-989-758-053-6

Quick Abstract

The increasing competition forces companies to use the capital more effectively and using suppliers which operate cheaper and with higher quality. Due to that, it is crucial to select the right suppliers. Supplier selection is a decision making problem that involves quantitative and non-quantitative, conflicting criteria. In The Linear Programming Technique for Multidimensional Analysis of Preference (LINMAP), all the decision data are known precisely or given as crisp values. But the uncertainty in the real life problems makes decision making more difficult. İn these situations, the complex situation varying with respect to decision makers can be solved by Fuzzy logic. Because of that, Fuzzy LINMAP has been used to solve the problem. The main aim of this study is to provide an analytical approach to decision makers for them to make objective decisions. Thus, supplier alternatives and selection criteria are determined. And a fuzzy LINMAP model is developed for supplier evaluation and selection of a company in automotive sector.

Related Ontology Subjects/Areas/Topics (21)

Artificial Intelligence	Fuzzy Systems	Neurotechnology, Electronics and Informatics
Biomedical Engineering	Health Engineering and Technology Applications	Pattern Recognition
Biomedical Signal Processing	Human-Computer Interaction	Physiological Computing Systems
Complex Fuzzy Systems	Industrial, Financial and Medical Applications	Sensor Networks
Computational Intelligence	Methodologies and Methods	Signal Processing
Computer-Supported Education	Neural Networks	Soft Computing
Domain Applications and Case Studies	Neurocomputing	Theory and Methods

Flatness based Control of a 2 DOF Single Link Flexible Joint Manipulator

Topics: Dynamical Systems Models and Methods; Robotics, Automation and Mechatronics

Authors: E. D. Markus ¹ , J. T. Agee ¹ , A. A. Jimoh ¹ , N. Tlale ² and B. Zafer ³

Affiliations: ¹ Tshwane University of Technology, South Africa , ² CSIR, South Africa , ³ Kocaeli University, Turkey

Digital Object Identifier: 10.5220/0004061104370442

Proceedings of SIMULTECH 2012, ISBN: 978-989-8565-20-4

Quick Abstract

The demand for high speed robotic manipulators with little or no vibrations has been a challenging research problem. In this paper, a position control for a 2 DOF single link flexible manipulator with joint elasticity is studied. It is shown that using the flatness control approach, faster response and less oscillations and overshoots can be achieved. The flat output of the linearized system is determined as the tip of the manipulator end effector. This output and a finite order of its derivatives is defined in terms of the input and states variables of the manipulator. Using the parameters of the output in flat space, a trajectory is planned and executed to test the effectiveness of the designed control.

Related Ontology Subjects/Areas/Topics (5)

Application Domains		Formal Methods		Simulation and Modeling
Dynamical Systems Models and Methods		Robotics, Automation and Mechatronics

Topic(s):

Title:
ISBN:
Year:
Acronym:
Subject:

Title:
Author:
Affiliation:
Subject:

Name:
Affiliation:
Country:
Conference:
Subject:

Name:
Country:
Subject:

Affiliation

Filter listed results, by