Thesis defense

thesis defense

Eray Üstün, Acoustic Source Localization Using Quaternion Fourier Transforms with Dırect-Path Dominance Test

M.S. Candidate: Eray Üstün
Program: Multimedia Informatics
Date: 23.01.2020 / 11:00
Place: A-212

Abstract: Acoustic source localization is an important and broadly researched topic since it can be used in robotics, scene analysis and surveillance fields. Acoustic scene analysis can be defined by acoustic pressure and particle velocity. The former one is a scalar value; whereas, the latter one is a vector quantity. Numerous approaches for DOA have been developed using those properties of acoustic waves. There is a 4-dimensional number system, called quaternions, which is an extension of classical complex number system. Quaternions are composed of one scalar component and 3-dimensional vector component. The quaternion Fourier transforms (QFTs) are useful in the sense that they can process components of multi-component signals concurrently. In this thesis, a novel signal model using quaternions is constructed and a novel approach for acoustic source localization are presented. The acoustical sound field is represented by quaternions and it is processed by using QFTs. Direct-path dominance (DPD) test is applied to the result of QFTs and localization task is performed by utilizing geometrical orientation of DOA estimation vectors.

Announcement Category

Emre Mülazımoğlu, The Role Of Visual Features In Text-Based Captchas: An fNIRS Study for Usable Security

M.S. Candidate: Emre Mülazımoğlu
Program: Cyber Security
Date: 15.01.2020 / 11:00
Place: A-108

Abstract: In order to mitigate dictionary attacks or similar undesirable automated attacks to information systems, developers mostly prefer using CAPTCHA challenges as Human Interactive Proofs (HIPs) to distinguish between human users and scripts. An appropriate use of CAPTCHA requires a setup balance between robustness and usability during the design of a challenge. The previous research reveals that most of the usability studies have used accuracy and response time as measurement criteria for quantitative analysis. The present study aims at applying optical neuroimaging techniques for the analysis of CAPTCHA design. In particular, fNIRS (Functional Near Infrared Spectroscopy) is a neuroimaging technique used for mental workload analysis by means of analyzing hemodynamic responses on brain. The present study reports an experimental investigation in which 25 participants solved a group of text-based CAPTCHA with various visual characteristics.

Announcement Category

Hatice Buket İkizoğlu, Exploring Hand Movement Dynamics During the Simon Task: A Mouse Tracking Study

Msc. Candidate: Hatice Buket İkizoğlu

Program: Cognitive Science

Date:  13.12.2019 14:00

Place: B-116

Abstract: Understanding the how the cognitive system processes information in real time is one of the key concerns for researchers in experimental psychology and cognitive science. Several measures and software tools have been proposed to explore different aspects of cognitive processing phenomena. One of these methods, mouse-tracking, helps researches to trace the participants’ hand movements during cognitive tasks which may be a reflection of their mental processes. Mouse-tracking method allows researchers to collect data about the dynamic unfolding of motor responses, and thus, provides additional insights about the process through which outcome-based measures such as reaction times or error rates are observed. Although the mental processes are treated as having discrete labels in traditional cognitive science, the new perspective indicates continuous processing of mind trajectories. Continuous and probabilistic representation of mental phenomena points out the continuous trajectory of the mind which travels through the set of possible brain states. Mouse Tracker’s real-time response data may shed light on continuous perspective of mind trajectories and temporal dynamics of cognitive processes. From this point of view the Simon Task which known as “stimulus-response compatibility effect” will be implemented with Mouse Tracker software. The collected multiple response alternatives will be analyzed in the light of continuous processing phenomena in an effort to shed further light into how the temporal dynamics of cognitive and motor processes contribute to the Simon effect.

Announcement Category

Babak Allahgholipour, Multi-Sensory Perception in Virtual World Animation Using Mcgurk Effect

Msc. Candidate: Babak Allahgholipour

Program: Multimedia Informatics

Date: 09.12.2019

Place: A-212

Abstract: Information about multisensory perception and brain cognition mechanism lead to a better virtual world creation. Brain store visual and auditory information to use in predicting future events. Image memory takes information from visual clues which during speech recognition happens by recording visual information and its auditory mappings. Based on this process, brain create prediction mechanism for future events. In speech recognition brain record sound and its visual presentation. When brain encounter difficulty in perception an auditory stimuli, it looks for visual mappings to guess the result. Considering the fact that some completely unrelated similarity may exist, some mapping may not lead to better understanding the world around. When previous information mislead the brain to understand the situation cognitive biases happen. McGurk Effect discusses how different lip movements affect auditory perception to get disparate information. Variety of factors influence amount of dependency on visual clues. Tracking speaking, especially in challenging environments, can be increased by visual clues such as face expressions, tongue and lips movements. Considering these elements in virtual world and game designing lead to improving believability. Perception processes, brain nature, designing factors, graphical elements, and visual structures are discussed to find out improved ways in designing realistic auditory and visual components.

Announcement Category

Melih Turhanlar, Detecting Turkish Phishing Attacks With Machine Learning Classifiers

Msc. Candidate: Melih Turhanlar

Program: Cybersecurity

Date:  06.11.2019 13:30

Place: A-108

Abstract: Phishing Attacks are a type of social engineering attack which aims to steal victim’s credit card numbers, credentials, and personal information by using victim’s curiosity and fear emotions. On these attacks, attackers send link with a text which triggers the victim's curiosity and fear emotions. If victim panic with the text inside attack vector and click the link, then victim connects to so-called web server or site which looks like a real web site but actually under control of the attacker. By filling the HTML forms inside the so-called web site, victims send their credentials unwittingly not to the real web site but actually to the attacker. In this study, we focused on phishing attacks with Turkish text in it. The data that we have used in that study contains not only Turkish phishing emails, but also Turkish phishing SMS and tweets that we gathered from the internet during the study. Focusing on Turkish texts inside the attacker’s attack vector, with Logistic Regression Synthetic Minority Over-Sampling Technique (0.923) F1 score has been achieved on our imbalanced dataset which we have created that can be used in other academic researches.

Announcement Category

SALVA DANESHGADEH ÇAKMAKÇI, A NOVEL ONLINE APPROACH TO DETECT DDOS ATTACKS USING MAHALANOBIS DISTANCE AND KERNEL-BASED LEARNING

PhD Candidate: SALVA DANESHGADEH ÇAKMAKÇI 
Program: Information Systems
Date: 22.11.2019 10:00
Place: Conference Hall 1

Abstract: Distributed denial-of-service (DDoS) attacks are constantly evolving as the computer and networking technologies and attackers’ motivations are changing. In recent years, several supervised DDoS detection algorithms have been proposed. However, these algorithms require a priori knowledge of the classes and cannot automatically adapt to the frequently changing network traffic trends. This emphasizes the need for the development of new DDoS detection mechanisms that target zero-day and sophisticated DDoS attacks. To fulfill this need, an online sequential DDoS detection scheme that is suitable for use with multivariate data was proposed. The proposed algorithm utilizes a kernel-based learning algorithm, the Mahalanobis distance, and a chi-square test. The algorithm is fully automated and does not require a pre-defined setting of any thresholds or baseline normal network traffic for training. Initially, four entropy-based and four statistical features were extracted from network flows as detection metrics per minute. Then, the kernel-based learning algorithm was employed to detect entropy-based input feature vectors that were suspected to be DDoS. This algorithm assumes no model for network traffic or DDoS; then, it constructs and adapts a Dictionary of features that approximately span the subspace of normal behavior. Every T minutes, the Mahalanobis distance between suspicious vectors and the distribution of Dictionary members is measured. Subsequently, the chi-square test is used to evaluate the Mahalanobis distance. The proposed DDoS detection scheme was applied to the CICIDS2017 dataset and the performance of the algorithm was measured using different performance metrics including accuracy, recall, precision and ROC-Curve. Finally, the results were compared with those by existing algorithms. It was demonstrated that the proposed online detection scheme outperforms almost all available DDoS classification algorithms with an offline learning process.

Announcement Category

Ali Mert Ertuğrul, Interpretable Spatio-Temporal Networks for Modeling and Forecasting Societal Events

PhD Candidate: Ali Mert Ertuğrul
Program: Information Systems
Date: 20.11.2019 15:30
Place: Conference Hall 1

Abstract: The relationships between individual activities and societal events (e.g. migrations, social movements) are complex due to the various social, temporal and spatial factors. Understanding such relationships in the context of various societal events such as street protests and opioid crisis, and forecasting these events is important as they have great impacts on public policies and supporting decision making of authorities. In this thesis, novel, spatio-temporal, deep neural networks are proposed (i) to forecast societal events and (ii) to help examine the relationships between societal events and their social and geographical contexts. The proposed models are designed to model the complex interactions between local (observed from within a location) and global (observed from all locations) activities by incorporating a new design of attentional networks. They are capable of forecasting the occurrence of future societal events and allow for interpreting what features, from which places, and how they contribute to event forecasting. Within the scope of this thesis, extensive experiments are conducted to evaluate the proposed networks on two different types of population-level societal events, namely social movements and opioid overdoses, with multiple datasets. The results indicate that the proposed models achieve superior forecasting performance than the compared methods. Also, they provide meaningful interpretations in terms of (i) what local and global activity features are more predictive, (ii) what locations have more salient contributions, and (iii) how these locations contribute to forecasting the subsequent events.

Announcement Category

Doğan Doğanay : Identifying Factors Affecting Auditors’ Adoption of Computer Assisted Audit Tools and Techniques (CAATTs): An Empirical Investigation

Msc. Candidate: Doğan Doğanay

Program: Information Systems

Date: 06.09.2019 11:00 a.m

Place: B-116

Abstract: Increasing usage of Information Technologies in organizations both private and public, audit activities has become more complicated for audit bodies. At this stage, Computer Assisted Audit Tools and Techniques (CAATTs) provide many advantages to auditors to carry out their tasks in an effective and efficient manner in such an environment and expansion of CAATTs usage plays an important role for auditors and organizations. In order to increase usage of CAATTs, it is critical to know what factors are significantly affecting the adoption decision. In this respect, the main objective of this paper is to reveal the factors affecting the Acceptance or Adoption of CAATTs by auditors. For this purpose, this study empirically explores the variables impacting use of CAATTs by Turkish auditors. As a result, a CAATTs adoption model is created in this study. In the scope of this study, firstly, studies related with the adoption of CAATTs were reviewed from 2000 to end of February 2019. This review gives information about past research on the field. At the end of the literature review, most significant factors affecting the CAATTs adoption are identified. Then, a technology adoption model and related hypotheses are proposed in the light of information derived from literature review. To test the hypotheses a quantitative method (questionnaire) is followed. Data is collected from auditors from Turkey. The model is tested using Structural Equation Modelling with Partial Least Squares (SEM-PLS). Inter-factor relationships are also introduced to the model after outcomes are obtained. At the end, the model's final version is developed and the most significant factors affecting the adoption of CAATTs by auditors are revealed.

 

Announcement Category

Mutlu Erhan : EVALUATION OF THE IOT DESIGNS AND PRODUCTS IN THE CONTEXT OF SECURITY AND PRIVACY

Msc. Candidate: Mutlu Erhan

Program: Information Systems

Date: 04.09.2019 16:15 p.m

Place: A-108

Abstract: Security and privacy issues in the Internet of Things (IoT) have received much attention in recent years because of the attacks, which have increased both in quantity and diversity. Many studies have been done to make the IoT ecosystem more secure, and these have managed to ease some risks partially by presenting security frameworks or basic standards. However; presented frameworks or standards have not been accepted by all the stakeholders in the IoT ecosystem and have not been able to provide solutions for design and evaluation. One way to decrease the risks posed by the vulnerabilities is to increase awareness of the stakeholders for security and privacy issues in the IoT system via providing simple, usable and enough protection skills, methods, standards and framework models in a design and evaluation environment.

Previous studies have analyzed reference framework models, presented security threats as a layered structure and managed to demonstrate the visibility of risks with a model of building blocks. However, besides the demonstration of the general security problems in the IoT stack, little attention was given to the generation of an evaluation environment and its usability. This study aims to present an environment, named as the Secure IoT Design and Evaluation Environment (SIDE), for IoT system developers to evaluate their products security risks against related vulnerabilities and to correct their deficits in the ecosystem, especially at the design phase. It was shown that the SIDE is practical and highly usable in identifying threats related to a design decision and evaluating the security of alternative solutions based on their known vulnerabilities.

Announcement Category

Fatma Mutlu : The Right Person to the Right Job: Developing a Two-Sided Matching Methodology Based on Real Employee Data

Msc. Candidate: Fatma Mutlu

Program: Information Systems

Date: 03.09.2019 14:00 p.m

Place: A-108

Abstract: Talented employees is one of the most important factors that carries companies to success in today's business world. However, making employees work in the right position so that it is compatible with their ability, nature and capacity is a much more important factor for success. Ignoring this situation poses an obstacle to work in an efficient and effective way for the companies. The objective of this thesis is to provide a methodology to match employees with the right position by considering both technical and behavioral competencies. In this study, both the needs of the employees and requirements of positions are taken into considerations and weights of them are calculated. Then a multi objective optimization model is developed to make both employee and position satisfactory degree the most. Results of this study are used for the purpose of achieving high job satisfaction and productivity by improving bilateral matching evaluation of both workers and positions. This study also may be used as guidance in the planning of businesses related training and development activities. To show the applicability and contribution of the methodology developed, it is validated using real life data.

Announcement Category

Pages

Subscribe to RSS - Thesis defense