2020 Spring - Senior Projects

A Nanonetworking Simulation Library

We design and implement a flexible, easy-to-use, high performance, GPU-accelerated nanonetworking simulation library.

Students: Kartal Kaan Bozdoğan

Randomized Gradient Adversarial Training

Randomized Gradient Step scales sign gradients pixel-wise in order to improve state of the art Projected Gradient Descent adversarial training to have higher diversity in adversarial generation thereby improve robustness and generalization of deep neural networks.

Students: Oğuz Kaan Yüksel

Advisors: İnci Meliha Baytaş

Explaining A Deep Learning Based Model for Spelling Correction of "de/da" Clitics in Turkish through Understanding Their Dependence to Various Regions of A Sentence

A model agnostic perturbation-based explanation approach is proposed for the spellchecker of Turkish “de/da” spelling errors.

Students: Alperen Değirmenci, Hasan Öztürk

Advisors: Suzan Üsküdarlı

A Security Bot to Automatically Detect Security Flaws in Web Applications

We developed a security suit that finds SSL/TLS vulnerabilities in networks and SQL injection vulnerabilities in websites.

Students: Yusuf Mert Bila, Seyfi Kutay Kılıç

Advisors: Fatma Başak Aydemir

A Software Traceability Tool

GUC is a software traceability tool that is initialized when project is created and can be updated during its development where users are able to see the relations between the architecture components.

Students: Fatih Maytalman, Omer Faruk Toptas

Advisors: Fatma Başak Aydemir

A Tool for Customized Searching of Research Articles (ASSORT)

Knowledge in the literature is increasing day by day.

Students: Zeynep Işık, Semih Akgül

Advisors: Suzan Üsküdarlı, H. Birkan Yılmaz

A Transcriptomics Analysis

Transcriptomics field studies human transcriptome, in other words human RNA.

Students: Ömer Faruk Özdemir

Advisors: Taflan Gündem

Adversarial Attacks on Time Series Data

Security vulnerabilities of state-of-art deep learning model for time series (LSTM) are exposed for regression and classification by adversarial attacks: Basic Iterative and Fast Gradient Sign Method.

Students: Afra Akbaş, Ceren Tahtasız

Advisors: İnci Meliha Baytaş

An API Testing Tool

This project is created for capturing the network flow between a mobile application and a REST API.

Students: Hasan Basri BALABAN, Burak YÜKSEL

Advisors: Alper Şen

An Approach for Reducing Return Rates for Clothing E-Commerce through Customized Size Recommendations

A size recommendation system for clothing e-commerce.

Students: Ahmet Yavuz Demir, Yunus Ege Saygılı

Advisors: Suzan Üsküdarlı, İnci Meliha Baytaş

Antibiotic Resistance Detection via Machine Learning

Developing visualization tools using neural network classifiers for novel antibiotic resistance detection to suggest potential mutations and to have a better understanding of the biological aspect of the predictions.

Students: İbrahim Kaplan, Onur Kılıçoğlu

Advisors: Arzucan Özgür

Autocompletion for Prefix-Abbreviated Input

This project implements and tests the performance of nested-trie indexing approach for autocompletion of prefix-abbreviated inputs.

Students: Muhammed Emin AYAR

Advisors: Taflan Gündem

Automated Curation of Labeled NER Datasets

This project implements a language-independent approach to generate named entity labeled datasets, rich with ambiguous terms, using Wikimedia.

Students: Barış Ege Sevgili, Alkım Ece Toprak

Advisors: Suzan Üsküdarlı

Autonomous Car

We developed autonomous car having 4 sensors and 4 wheels using LPC4088.

Students: Sercan Ersoy

Advisors: Arda Yurdakul

Autonomous Vehicle Mobility within 5G Infrastructure

p.

Students:

Advisors: Cem Ersoy, Atay Özgövde

Bilateral Molecular Communication

Simulation that allows two different cell sets to send molecules to each other and communicate with each other in a molecular way.

Students: Esra Yılmaz

Advisors: Tuna Tuğcu

Binârî: the Artificial Ghazalkhwan

This project tries to generate meaningful, formally correct Ottoman ghazal poems using deep learning, natural language processing techniques with the data created by the Ottoman Text Archive Project.

Students: Galip Ümit Yolcu

Advisors: Tunga Güngör

Building the Base of a Multi-Purpose Social Robot

This project describes the mechanical, electrical, and software steps of building the base of a multi- purpose social robot that is designed to blend in and aid to humans by navigating similar to them.

Students: Bartu Ören

Advisors: H. Levent Akın

Customized Academic Job Finding Application

Aim of the project is to ease the process of finding academic jobs.

Students: Baturalp Yörük, Mehmet Berk Kemaloğlu, Muhammed Bera Kaya

Advisors: H. Birkan Yılmaz

Decentralized and Ontology Based Purposeful Online Community Application Framework

The goal is to build a knowledge-centric online community model that supports the specification of a privacy concerned, processable and analyzable data producing application by non-tech savvy users.

Students: Serkan Özel

Advisors: Suzan Üsküdarlı

Deep Learning Applications for Shopping Industry

For CMPE492 Senior Project, we focused on the modern applications of Deep Learning and came up with two ideas for shopping industry.

Students: Ali Meriç Deşer, Mete Han Kurt

Advisors: Fikret Gürgen

Deep Learning Techniques for Stock Price Prediction

We propose a deep model comprising LSTM and CNN architectures for Stock Price Prediction.

Students: Efe Önal, Barış Zöngür

Advisors: İnci Meliha Baytaş

Deep Transfer Learning

We study transfer learning, which aims to adapt previously gained knowledge to small scale problems.

Students: Furkan Kadıoğlu

Advisors: İnci Meliha Baytaş

Design & Development of a Distributed Ledger Enabling Cross-Border e-Procurement

In this project we will build a system using Distributed Ledger Technology that enables Cross-Border e-Procurement.

Students: Ahmad Jawed ZAFAR

Advisors: Fatih Alagöz

Design and Implementation of a Distributed Ledger Enabling Cross-border E-procurement

We design and implement more secure and easily managed, distributed ledger e-Procurement network using blockchain, alternative to in-use e-Procurement system PEPPOL managed by a central authority.

Students: Mahmut Uzunpostalci, İbrahim Kamacı

Advisors: Fatih Alagöz

Distributed Ledger Technology-based Vulnerability Database Using Machine Learning Methods

There exists many vulnerability databases but these databases are generally centralized which arises a huge controversy with being a secure database because centralized databases contain some security problems such as SQL injection attacks.

Students: Hilal Mente

Advisors: Fatih Alagöz

Experimental Analysis of Persuasive Effects of Emotions on a Social Robot

We investigated the persuasive effects of emotions on a social robot through an online HRI experiment that is a web based survey on COVID-19 awareness conducted by a Nao robot.

Students: Yağmur Kahyaoğlu, Elif Çalışkan

Advisors: H. Levent Akın

Human Motion Prediction with Graph Neural Networks

Modelling the full-body motion and predicting the movement of an articulated human by exploiting the underlying graph structure and using graph neural networks.

Students: Irmak Guzey

Advisors: Emre Uğur

IDE-Independent Plug-in for Programming Development

Project aims to convert the plug-in project prepared in Eclipse into Intellij.

Students: Cumhur Kılıç

Advisors: Tuna Tuğcu

Implicit Biometric Authentication with Smartwatches

Today’s technology enables biometric data to be obtained through wearable devices.

Students: Yağmur Ceren Dardağan

Advisors: Cem Ersoy

Improving Robustness of Deep Learning (Evolutionary Computation Adversarial Attack)

We propose a new Adversarial Attack combining Projected Gradient Descent and Evolutionary Computation.

Students: Samet Demir

Advisors: Alper Şen

Integrating Goal Models to a Model Generation Pipeline

Our aim is to be able to visualize goal models automatically from JSON data in an interactive and editable manner with user-friendly and an aesthetic interface.

Students: Ebru Zorlu, Gamze Gülbahar

Advisors: Fatma Başak Aydemir

Low Latency Trading Platform

We introduce a Low Latency Trading Platform which allows the trader to react to sudden changes in financial markets in nanoseconds.

Students: Mehmet Altay Ince, Emirhan Yasin Cetin

Advisors: Arda Yurdakul

Mobile Test Automation Project

Mobile Test Automation Framework is a project that aims to automate the testing process for mobile applications by collecting data based on real-world usages and then by feeding this collected data to a test generation engine that is based on machine learning.

Students: Fatih İver, Metin Dumandağ

Advisors: Alper Şen

Performance Evaluation of Edge Computing Architectures on Autonomous Vehicles within 5G Infrastructure

The idea of building a fully working autonomous vehicle is closely related to a proper working Internet of Vehicles.

Students: Ege Başural

Advisors: Cem Ersoy

Performance of Multi-threaded WebAssembly Applications on Browsers

WebAssembly multithreaded performance in browsers and on traditional hardware is compared, we found targeting WebAssembly is a valid choice for various multithreaded applications.

Students: Egemen Göl

Advisors: Can Özturan

Predicting Whole Body Motion Trajectories using Conditional Neural Movement Primitives

This study offers a new approach to whole body human motion prediction problem using a state-of-the-art robot learning from demonstration framework,the Conditional Neural Movement Primitives.

Students: Mehmet Hakan Kurtoğlu

Advisors: Emre Uğur

Rare Disease Detection with Meta-Learning

Our problem is rare disease detection which can be posed as a few- shot classification problem from machine learning point of view.

Students: Burak Çuhadar, Özgür Solak

Advisors: İnci Meliha Baytaş

Reducing Memory Usage with a Data-Aware Index Structure

Indexing structures might be responsible for a high percentage of database memory consumption.

Students: Aysu Sayın, Alican Şafak Özdek

Advisors: Taflan Gündem

Resilience in Deep Neural Networks

Are the deep learning models resilient enough to be used in real life?

Students: Sadullah Gültekin

Advisors: Alper Şen

Restful Spring Initializr

In modern Spring Boot applications, there are parts that are common in many projects.

Students: Ahmet Gedemenli

Advisors: Fatma Başak Aydemir

Salt Detection with Deep Learning

Interpreting seismic images has important role in lots of industry areas it is highly used by people whose profession is geophysics.

Students: Harun Acar

Advisors: Fikret Gürgen

Scalable, Self-Hosted Testing Suite for Mobile Applications

Easy-to-use, scalable, self-hosted testing platform for Android and iOS applications.

Students: Bahadır Hocamoğlu

Advisors: Alper Şen

Security Cam With Image Processing Report

Security cam project aims to make homes safer and check the people who rings the bell at the door.

Students: Mustafa Şişik

Advisors: H. Birkan Yılmaz

Self-training with Noisy Student

I used a self-training method to improve classification accuracy on CIFAR-10 and MNIST datasets.

Students: Hasan Yaman

Advisors: Lale Akarun

Semi-Automatic Segmentation and Sign Discovery

Our project is semi-automatic segmentation of sentences & words and automatic finding exact starting and ending frames of sign words' in these segments in SEBEDER and MeineDGS Sign Language dataset.

Students: Berke Esmer, Cemal Aytekin

Advisors: Lale Akarun

Sign Language Avatar Generation

Little data containing sign language leads to inadequate research.

Students: Burak İkan Yıldız, Gürkan Demir

Advisors: Lale Akarun

Simulation of Reception Using Two Different Molecule Types

Molecular communication is a very useful way to carry data using nanodevices.

Students: Neval Tüllük

Advisors: Tuna Tuğcu

Simulator Performance Improvements for Cubic Receiver

Molecular communication is relatively new research field that is inspired from biological cell communication.

Students: Hilal Demir

Advisors: H. Birkan Yılmaz

SliceSim2: A Load Balancing Simulation Suite for 5G Network Slicing

Network slicing allows multiple virtual networks to run on a single hardware system.

Students: Enes Turan ÖZCAN, Ufuk YILMAZ

Advisors: Tuna Tuğcu

Smartwatch Based Continuous Authentication with Deep Learning Methods

Security of the private user data stored in cloud services and IoT devices is a serious concern.

Students: Dilruba Köse, Furkan Aydar

Advisors: Cem Ersoy

SONAR: Social Network Analysis of Researchers

SONAR aims to identify the key researchers in a domain of study through creating and analyzing author-centric social networks.

Students: Halit Ozsoy, Melih Demir

Advisors: Suzan Üsküdarlı, H. Birkan Yılmaz

TeachingCodes Course Content Editor

This is a web based markdown plain text editor with customizable additional syntax and side by side HTML preview for the purpose of being used with TeachingCodes Eclipse plugin.

Students: Doğa Yüksel

Advisors: Haluk O. Bingöl

Training Walking Data with CNMP and ACNMP

In order to make robots have different skills, learning relations between complex temporal sensorimotor values with external parameters and goals is essential.

Students: Abdullah Enes Öncü

Advisors: Emre Uğur

Transaction Gas Analysis in Ethereum Blockchain

Ethereum blockchain works on Gas.

Students: Imran Siddiq

Advisors: Arda Yurdakul

Turkish – English Machine Translation System

We organized a Turkish-English parallel corpus that can be used with machine translation applications.

Students: İrem Uğuz, Sümeyra Yılmaz

Advisors: Tunga Güngör

Twitter Sentiment Analysis for Boğaziçi University

Twitter Sentiment Analysis for Boğaziçi University using advanced text mining techniques (Naive Bayes, Word2Vec and BERT) to analyze the sentiment of the text in the form of positive, negative and neutral.

Students: Oğuzhan Kırlar, Batuhan Enes Çağlayan

Advisors: Arzucan Özgür

User Segmentation and Propensity Score Matching Using Clickstream Data

Our aim is to detect users propensity to purchase products by analyzing their transactions and recommend the products they will possibly buy.

Students: Hatice Melike Ecevit, Merve İlik

Advisors: İnci Meliha Baytaş

Using Machine Learning for Testing Mobile Applications

Applying deep reinforcement learning methods for tasks whose rewards are specified by a linear temporal logic formula; for the purpose of automatic test case generation for Android GUI applications

Students: Ahmet Selim Karaduman

Advisors: Alper Şen

Verifying Text Summaries of Relational Datasets

As a result of widespread use of internet as an outlet to reach the information, people are struggling with a bizarre number of false claims and half-truths.

Students: Kemal Duru

Advisors: Taflan Gündem

Web-Based Quality Defect Detector for Turkish User Stories

This project aims detecting quality defects for Turkish user stories, making suggestions for the detected defects and developing a web application which makes user stories more reliable and error-free.

Students: Fahri Can Şanlı, İrem Üstünboyacıoğlu

Advisors: Fatma Başak Aydemir

Wrecked Plane Localization Simulator

Molecular communication is drawing attention recently, and it has its own benefits compared to other methods.

Students: Kaan Özgen

Advisors: H. Birkan Yılmaz

Completed Projects - 2020 Spring