Project Title: Simulating Quantum Computing using CUDA

Project Description:Computational requirements are in huge demand in todays’ applications and this reality is gaining its importance every day. We need not only to make a single process faster, but also achieve many tasks at the same time with a much faster rate. One extra ordinary solution proposal to our ever increasing computational needs is quantum computing.
Rather than store information using bits represented by 0s or 1s as conventional digital computers do, quantum computers use quantum bits, or qubits, to encode information as 0s, 1s, or both at the same time. This superposition of states—along with the other quantum mechanical phenomena of entanglement and tunneling—enables quantum computers to manipulate enormous combinations of states at once. To speed computation, quantum computers tap directly into an unimaginably vast fabric of reality—the strange and counterintuitive world of quantum mechanics.
In this project, we will create a quantum computer simulation using Compute Unified Device Architecture (CUDA) to take advantage of the high computing capacity of the Graphics Processing Units (GPU), and then we will try to observe foundational quantum computation samples on this simulator.

Requirements and Additional Info:CUDA, parallel programming concepts

Contact: If interested, please send your resume and your letter of intention to Adnan Özsoy

Project Title: Mining COSMIC functional size measurement patterns from software code

Project Description:Functional size (FS) of a software program is the (transaction-based) functionality offered by the program to its users. The difference of FS from Lines of Code (LOC) metric is that FS can be measured as independent of programming language, paradigm, or environment. FS provides a base for comparing the sizes of various software programs, and also for estimating effort and cost required to developed the programs. It is very critical for software projects that FS is calculated automatically from software code, like LOC metric, however; the process of automation is not straightforward, and therefore is an emerging research problem. In this project, you are required to work on automatic calculation of functional size from software code. More specifically, you are expected to mine (identify and list) already defined patterns of COSMIC functional size measurement from software applications developed in Java. (For more information on COSMIC Functional Size Measurement, please visit http://cosmic-sizing.org/)

Requirements and Additional Info: BBM102, BBM104, BBM201, BBM202, BBM203, BBM204

Contact: If interested, please send your resume and your letter of intention to Ayça Tarhan

Project Title: Multilingual Video Captioning Dataset

Project Description: Integrated vision and language research has attracted increasing attention in different areas of AI research. However, the focus in the community is currently on mono-lingual setting and specifically in English. In this project, we will create a benchmark multi-modal video dataset for multilingual vision and language research. Dataset will include short (4-5 secs.) video clips extracted from the movies having closed captioned descriptions. Each video clip will be associated with sentence descriptions in both English and Turkish, describing the main activities and the objects within the scene. All these descriptions should be manually aligned within each other and to the corresponding video clip.

Requirements and Additional Info: Candidates should have strong programming skills and mathematical background.

Contact: If interested, please send your resume and your letter of intention to Erkut Erdem and Aykut Erdem

Project Title: Parallel Corpora Collector

Project Description:Most machine translation algorithms depend on parallel corpora. A parallel corpus is simply a list of sentence pairs where each pair contains a sentence in source language and its corresponding translation in target language. Using these sentence-level aligned corpora, machine translation algorithms learn how to translate one language to other. Unfortunately parallel corpora is not readily available for many language pairs. For example it is not easy to find parallel corpora for Turkish.
On the other hand most international company web-sites serve their content in multiple languages. Crawling these web-sites to create a parallel corpora can be a good strategy for many low-resource languages. This project aims to build crawler tools to extract translated contents from the Web. This task can be considered in multiple phases where in the first phase given a list of example URL addresses, learn URL and HTML patterns to extract translated content and align the sentences. After the successful completion of phase 1, by giving a list of few example URL pages the rest of the data will be extracted automatically. In the second phase, automatic identification of translated pages will be attacked, so that the only input for the tool will be some domain names.

Requirements and Additional Info: Candidates should have strong programming skills and mathematical background.

Contact: If interested, please send your resume and your letter of intention to Gönenç Ercan

Project Title: High Level Synthesis (HLS) Tool

Project Description:An HLS tool is the system level synthesis part of the electronic design automation flow. It takes the application as a high level code and determines the RTL (Register Transfer Level) of the pplication that will be implemented as hardware. This project aims to create an HLS tool that schedules the operations of a given application and allocates resources to them. Student will implement ASAP, ALAP, and list scheduling algorithms.
An important note: There is no real hardware implementation in the project. Basic digital design knowledge is enough to understand the details of the problem.

Requirements and Additional Info:Good programming skills and digital design knowledge.

Contact: If interested, please send your resume and your letter of intention to Suleyman Tosun

Project Title: Evaluating the Effects of Voltage Scaling on GPGPUs

Project Description:Dynamic voltage scaling is a power management technique in computer architecture, where the voltage used in a component is increased or decreased, depending upon circumstances. Dynamic voltage scaling to increase voltage is known as overvolting; dynamic voltage scaling to decrease voltage is known as undervolting. Undervolting is done in order to conserve power, particularly in laptops and other mobile devices, where energy comes from a battery and thus is limited, or in rare cases, to increase reliability.
In this project, student will simulate parallel code on GPGPUs. Some portion of the code will run under low voltages to minimize the power consumption. The student will investigate the tradeoff between the power and performance of the simulated application.
For this purpose, NVIDIA CUDA framework and GPGPU-sim will be used to run the experiments.

Requirements and Additional Info:Good knowledge of parallel programming and concepts.

Contact: If interested, please send your resume and your letter of intention to Adnan Ozsoy, Suleyman Tosun

Project Title: Detection of Mobile Graywares

Project Description:Graywares are applications which are not fully malicious but might present security risks for the users. In this project, we aim to detect graywares on the official Android Market. We especially aim to focus on impostors that impersonate other popular apps to gain installation, and misrepresentators that falsely claim to provide functionality to users. In this project, we are going to use NLP techniques for differentiating graywares from benign applications on the market.

Requirements and Additional Info: Good programming skills are required.

Contact: If interested, please send your resume and your letter of intention to Sevil Şen and Burcu Can

Project Title: Analysis of Android Malwares on Alternative App Stores

Project Description:In this project, we aim to analyse malwares on alternative app stores. The applications from alternative app stores other than GooglePlay will be downloaded, and analysed statically and dynamically. Both the code and the behaviour of the application will be analysed in order to label downloaded applications as malwares.

Requirements and Additional Info: The candidates should be good at programming in Java. The proficient users of Unix/Linux systems will be preferred.

Contact: If interested, please send your resume and your letter of intention to Sevil Şen

Project Title: a Healthy Portion of VR

Project Description:We will look into ways of creating Virtual Reality (VR) applications/games to be used in the treatment of eating disorders and body-image problems. During the project, we will be collaborating with the faculty members from various departments of Hacettepe University Faculty of Medicine. You may take a role in developing the VR interface, creating the contents of VR scenes, integration of various VR-related technologies (Oculus Rift, Kinect Sensor etc.) with the VR applications, or designing and programming machine-learning methods to be embedded into the VR applications.

Requirements and Additional Info: Familiarity with Unity game-engine is a must. Also, experience in 3D modeling is a big plus for content-creator positions.

Contact: If interested, please send your resume and your letter of intention to Ufuk Celikcan

Project Title: VR-Fit

Project Description:We will look into ways of creating Virtual Reality (VR) applications/games to be used in (a) therapy of physical injuries and (b) general physical exercise. We will be collaborating with the faculty members from various departments of Hacettepe University Faculty of Medicine. Current ongoing tracks in which you can take a part include (1) motivational exercise programs for overweight teenagers (2) physical rehabilitation of shoulder injuries and knee injuries. More tracks are coming soon. You may take a role in developing VR interfaces, creating the contents of VR scenes, integration of various VR-related technologies (Oculus Rift, Kinect Sensor, Leap Motion Controller, Wii-Fit Board etc.) with the VR applications, or designing and programming machine-learning methods to be embedded into the VR applications.

Requirements and Additional Info: Familiarity with Unity game-engine is a must. Also, experience in 3D modeling is a big plus for content-creator positions.

Contact: If interested, please send your resume and your letter of intention to Ufuk Celikcan

Project Title: Virtual Reality in Rehabilitation and Therapy

Project Description:We will look into ways of using Virtual Reality (VR) applications/games in (a) rehabilitation of psychological disorders ( b) therapy of physical injuries and (c) general physical exercise. We will be collaborating with the faculty members from various departments from Hacettepe University Faculty of Medicine. Current ongoing tracks where you can take a part in include (1) treatment of eating disorders and body-image problems (2) treatment of performance anxiety in nation-wide standardized tests (such as YGS and LYS) (3) motivational exercise programs for obese teenagers (4) physical rehabilitation of shoulder and knee injuries (and more tracks are coming soon). You may take a role in developing VR interfaces, creating the contents of VR scenes, integration of various VR technologies (Oculus Rift, Kinect Sensor, Leap Motion Controller, Wii-Fit Board etc.) with the VR applications, or designing and programming machine-learning methods to be embedded in the VR applications.

Requirements and Additional Info: Familiarity with Unity is a must. Also experience in 3D modeling is a big plus for content-creators.

Contact: If interested, please send your resume and your letter of intention to Ufuk Celikcan

Project Title: Automated Model Driven Testing Framework – For Model Driven Development

Project Description:Software testing is a critical activity in software development life cycle. Manual testing is quite time consuming and also prone to errors. In order to reduce the risks and costs of manual testing, an integrated testing framework is needed. In this project we propose a framework that will offer an automated test architecture for model based systems. This framework will provide an integrated environment for test planning, test development, test execution and test analysis. In test planning process, the functional aspects of the system to be tested are identified. To be more specifically, detailed test cases are determined for the models extracted from the formal specifications (namely, requirements) of the system under test (SUT). By this way, requirements and test cases traceability is ensured. In test development phase, executable/physical test cases are designed and created for the models in the SUT. In test execution process, testing framework will provide a means of running the executable test cases generated in test development phase. In test analysis process, the success and failure of test cases will be monitored and reported to the developer/tester.

Requirements and Additional Info: Proficiency in a programming language e.g. C++ or .NET C# familiarity with model driven architecture, knowledge on software testing life cycle is a big plus, knowledge on standardized modeling language (UML).

Contact: If interested, please send your resume and your letter of intention to Adnan ÖZSOY, Aylin HATİP İPEK (company mentor)

Project Title: Transforming Street-level Images through Adversarial Learning

Project Description:Generative adversarial networks (GANs), with its many variants, are among the most popular topics in deep learning research today. These deep architectures have shown great promise in modelling natural images and synthesizing novel and realistic looking images from scratch. In this project, by specifically focusing on street-level imagery, we will explore the use of recently proposed GAN models to translate street images into a related target domain. This mapping process will be carried out by considering different visual transformation tasks, which will allow us to augment existing street image datasets with additional forms of annotations.

Requirements and Additional Info: Candidates should have strong programming skills and solid machine learning background.

Contact: If interested, please send your resume and your letter of intention to Aykut Erdem, Erkut Erdem

Project Title: mrb, nslsn? Text normalization for Turkish

Project Description:Length restrictions on some platforms like SMS or Twitter led to a new bad formed language, where instead of writing "merhaba" people started to write "mrb". Mapping these words to their correct form is crucial for many text mining applications. One possible approach to this problem is creating rule-based systems, for example removal of vowels. However this approach is limited as new forms appear each day. Another approach to this task is by observing the context of Out of vocabulary words and their corresponding In Vocabulary words to automatically find a mapping between them. In this project you will use context based vector representations of words such as Word2Vec or GloVE and solve text normalization problem.

Requirements and Additional Info: Data structures, good programming skills

Contact: If interested, please send your resume and your letter of intention to Gönenç Ercan