Student Research Opportunities
You're never too young to do research!
The professors of Ying Wu College of Computing are always looking for students to help them with their research.
Don't worry about "not being ready" for research. We have a 9th grade high school student taking part in Social Media research. Don't even worry if you are not sure what "doing research" means. You will learn it -- by doing it.
What you need is working knowledge of one common programming language (for example, Java, C, C++, Python, JavaScript or R). Some students even learn programming on their own, without taking a class.
You need one more thing: A good work attitude. Be on time for meetings, reply to email messages of your professor, tell your professor if you run into problems (including private problems) or if you don't understand your assignments, do what your professors assigns you, and make a little progress every week.
Why do research?
We professors do research because it is fun. Hopefully, you will also feel the fun and excitement of doing something new that nobody has ever done before.
Even more important, you can put your research on your resume. You can talk about it at job interviews and impress your interviewer, and it will prepare you for jobs where you are expected to do research. That widens the range of possible jobs you can take.
Academic Credit or Payment?
Projects can be done for payment (if funding is available) or for academic credit.
Contact the supervisor for more details and the timeline.
Academic credit can be obtained by enrolling in one of the following courses:
- CS 488 or IS 488 (Independent study for Undergraduates)
- CS 725 or IS 725 (Graduate Research 1)
- CS 726 or IS 726 (Graduate Research 2)
- CS 700B or IS 700B (MS Project - 3 hours)
- CS 701B or IS 701B (MS Thesis - 6 hours)
Projects
Participatory Learning: Revolutionizing Learning by Putting the Students in Charge
Computing the Global Minimum of a Continuous Function by Domain Subdivision
Building High Performance Cloud Infrastructures
Applications of AI in Digital Games and Creative Computing
Improving Smartphone Security and Reliability
Visualizing Complex Medical Terminologies
Dimensionality and Scalability Issues in High Dimensional Spaces
Analysis, Exploration and Visualization of Big Data for Traffic Congestion and Traveling Behavior Prediction
DeepPrivate - Differential Privacy Preservation in Deep Learning under Model Attacks
High-Assurance Open-Source Software for Cybersecurity and Cryptography
Blockchain Applications
NetExplorer
Deep Learning for DNA Sequence Pattern Recognition
Quantitative Stock Selection and Trading
Health Technologies and User Experience
Understanding Livestreaming and eSports
Write-and-Learn: Promoting Meaningful Learning Through Concept Map-Based Formative Feedback On Writing Assignments
Simulating Spatial Meme Diffusion
Building a Framework of Software and Hardware for Real-time Predictive Analytics on Social Networks
Interactive Mix Reality Platform
Science and Engineering in Program Repair
Mining Big Data through Deep Learning
AI Deep Learning in Medical Image Analysis and Other Applications
Participatory Learning: Revolutionizing Learning by Putting the Students in Charge
Supervisor: Michael Bieber
Email: michael.p.bieber@njit.edu
Homepage: http://web.njit.edu/~bieber/
Description: Students may be motivated to deeper learning if they play an active role in all parts of the problem life-cycle. The framework for this research project relies on students to make up the problems, solve them, grade solutions and dispute their grades - all with the support of an online system allowing them to read everything their peers have done. Subjects for researching include: Will this work for homework, labs, projects, quizzes, final exams? Will it work for essays, science labs, math problems, computer programs? What about anonymity and group work? What would it take for students and instructors to trust this approach? Projects could design experiments for these aspects, or design new features such as group support and gaming features, or build in solid security features.
Prerequisites: For security features, Javascript and Node.js experience is necessary. For design and development research, JavaScript and SQL experience is necessary. No programming but a keen interest is required for user experience testing or experimental design.
Computing the Global Minimum of a Continuous Function by Domain Subdivision
Supervisor: James Calvin and Craig Gotsman
Email: james.m.calvin@njit.edu
Description: Many applications in computer science and engineering require the solution of an opti- mization problem, namely the minimization of some cost function, which would imply that the solution to the problem is the best possible. Examples include:
In constructing artificial neural networks, it is desirable to choose a set of weights to minimize training error.
In image registration, the goal is to align images (for example medical images taken of the same body area at different times) while minimizing a misalignment cost.
In data clustering, the goal is to partition data points into groups that are similar.
The goal is to find the global minimum of the cost function. Unfortunately, most cost functions have multiple local minima, and standard optimization algorithms are capable of finding only a local minimum of the cost, which may be quite worse than the global minimum. This project addresses the important problem of finding a global minimum of a continuous cost function by a suitable software algorithm.
The approach we propose to compute a global minimum is based on “searching” for the global minimum in the function domain by adaptively subdividing the domain, narrowing down the region of the domain where the global minimum is to be found. Thus, the subdivision becomes finer where the cost function is smallest. The figure on the right below depicts a subdivision into triangles for the cost function depicted on the left. A different subdivision scheme could be based on rectangles and recursive partitioning.
The cost of refining the subdivisions can grow rapidly with the dimension of the do- main. The purpose of this project is to develop efficient data structures and algorithms for subdivision refinement, and explore their use in optimization algorithms. The project will also investigate the application of the algorithms to different computing problems, possibly the ones listed above. The project will involve sophisticated software development in an object-oriented language.
Prerequisites: Experience in programming, basic knowledge in data structures and algorithms at the level of CS435.
Building High Performance Cloud Infrastructures
Supervisor: Xiaoning Ding
Email: xioning.ding@njit.edu
Homepage: http://web.njit.edu/~dingxn/
Description: The project improves virtualization to allow programs in virtual machines to make efficient utilization of hardware resources, such as multicore processors and memory devices. The main approach is to create expressive interfaces to hardware resources, with which programs in virtual machines can obtain better knowledge and gain more control over the hardware resources to optimize performance. Three main tasks in the project are 1) creating expressive interfaces, 2) developing programs that can optimize performance utilizing the expressive interfaces, and 3) testing the performance of the programs.
Prerequisites: For students interested in task 1, skills of modifying and building Linux kernel are required. Students interested in task 2 must be able to write multi-threaded C/C++ programs on Linux systems. Students interested in task 3 are required to write scripts on Linux systems.
Applications of AI in Digital Games and Creative Computing
Supervisor: Amy Hoover
Email: amy.k.hoover@njit.edu
Homepage: http://amykhoover.com/
Description: I work on AI systems that together with humans, collaboratively make games, music, sound, or art. My research acknowledges that humans and computers excel in different areas of the creative process, and draws on these unique talents to build systems that harness the power of each. My work develops methods for designing games and facilitating human-computer collaborations that focus on solving problems in digital and creative domains. I am looking for motivated students interested in these domains. Programming competence in C#, Java, or Python is important.
Improving Smartphone Security and Reliability
Supervisor: Iulian Neamtiu
Email: iulian.neamtiu@njit.edu
Homepage: https://web.njit.edu/~ineamtiu/
Description: Users are increasingly relying on smartphones, hence concerns such as mobile app security, privacy, and correctness have become increasingly pressing. Prof. Neamtiu’s group is working on filling thisgap through tools that permit a wide range of software analysis for the Android smartphone platform, e.g., static analysis, dynamic analysis, record-and-replay or network traffic profiling. Our tools aim to analyze substantial, widely-popular apps (e.g., Yelp, Facebook) running directly on smartphones, and without requiring access to the app's source code. Our results include finding bugs in popular apps, high-fidelity record-and-replay, exposing risky URLs, self-healing apps, etc.
Prerequisites: Experience with Android or iOS development AND strong programing skills.
Visualizing Complex Medical Terminologies
Supervisors: Yehoshua Perl, James Geller and Christopher Ochs
Email: james.geller@njit.edu
Homepages: http://www.cs.njit.edu/~perl/; http://web.njit.edu/~geller/
Description: Biomedical ontologies are large and complex knowledge representation systems. We have developed a software system called the Ontology Abstraction Framework (OAF) to create, visualize, and explore summaries of ontologies called abstraction networks. In our research ewe use abstraction networks to support comprehension of ontology structure, ontology quality assurance, and ontology change analysis. The OAFis composed of several modules, each of which enables the summarization of a different aspect of an ontology's structure. In a current project we are extending the OAF to support "Live Abstraction Networks," which summarize an ontology as a user is editing it.
Prerequisites: Students interested in working on the OAF project should have experience designing and developing software projects in Java. Experience with Swing, threads, JSON APIs, and Lambda are recommended. A strong background in CS theory, with a thorough understanding of trees and graphs, is required. Experience with Java IDEs, debuggers, profilers, and Git are a plus.
Dimensionality and Scalability Issues in High Dimensional Spaces
Email. vincent.oria@njit.edu
Homepage: https://web.njit.edu/~oria/index.htm
Description: For many fundamental operations in the areas of search and retrieval, data mining, machine learning, multimedia, recommendation systems, and bioinformatics, the efficiency and effectiveness of implementations depends crucially on the interplay between measures of data similarity and the features by which data objects are represented. When the number of features (the data dimensionality) is high, the discriminative ability of similarity measures diminishes to the point where methods that depend on them lose their effectiveness. We are investigating effective feature selection techniques, and their application to search and clustering.
Analysis, Exploration and Visualization of Big Data for Traffic Congestion and Traveling Behavior Prediction
Supervisor: Hai Phan
Email: hai.phan@njit.edu
Description: The long-term effects of traffic congestion may cost the U.S. government and American taxpayers hundreds of billions of dollars annually. Emissions of gases from billions of gallons of fuel lost in gridlock cause global warming and environmental degradation. Long commutes are associated with lower fitness levels, higher weight, and higher blood pressure, all of which are strong predictors of heart disease, diabetes, and different types of cancer. To slow or even reverse the trend of growing gridlock, accurately predicting traffic congestion and traveling behavior is desirable. It leads to more effective investment decisions for transportation improvements, which affect safety, environmental quality, economic development, quality of life, and lower health risks. This project aims at developing innovative solutions using cutting-edge technologiessuch as Internet of Things (IoT) and deep learning to analyze, explore, and visualize big data for traffic congestion and traveling behavior prediction. This project takes an integrated approach to (1) modeling andrepresenting dynamic traffic flow over time; (2) contextually predicting traffic congestion; (3) modeling and forecasting traffic influence networks (TINs), in which traffic conditions in one location may affect traffic conditions in other locations; (4) predicting traveling behavior, including changes in routes and departing time, in both long and short terms and (5) querying and visualizing large-scale urban transportation data.
Prerequisites: Strong in Python and have a background about data mining, machine learning. Students will have opportunities to work with cutting edge-technologies such as big data and deep learning in urban data science. Students are expected to implement and deploy practical tools to predict and visualize traffic congestion and human behavior in traveling.
DeepPrivate - Differential Privacy Preservation in Deep Learning under Model Attacks
Supervisor: Hai Phan
Email: hai.phan@njit.edu
Description: Today, the remarkable development of deep learning in medicine and healthcare domains presents obvious privacy issues, when deep neural networks are built based on patients' personal and highly sensitive data, e.g., clinical records, user profiles, biomedical images, etc. To convince individuals to allow that their data be included in deep learning projects, principled and rigorous privacy guarantees must be provided. However, no deep learning techniques have yet been developed that incorporate privacy protection against model attacks, in which adversaries can use released deep learning models to infer sensitive information from the data. In clinical trials, such lack of protection and efficacy may put patient data at high risk and expose health care providers to legal action based on HIPAA/HITECH law. This project will develop a mechanism, called "DeepPrivate," for privacy preservation in deep learning under model attacks. The PIs' mechanism will offer strong privacy protections for data used in deep learning. To put the DeepPrivateframework to work, fundamental challenges in differential privacy preservation in deep learning under model attacks need to be synergistically overcome. Consequently, this project will advance the state-of-the-art in key questions: (1) The framework design to preserve differential privacy in various types of deep neural networks; (2) The utility maximization of the models; (3) The potential model attacks in deep neural networks under differential privacy; (4) The information disclosure prevention approaches ; and (5) The multiparty computation protocols in deep learning under model attacks.
Prerequisites: Strong in Python and have a background about data mining and machine learning. Students will have opportunities to work with cutting edge-technologies such as deep learning and security and privacy in data science. Students are expected to implement and deploy practical tools in security and privacy in deep learning.
High-Assurance Open-Source Software for Cybersecurity and Cryptography
Supervisor: Kurt Rohloff
Email: kurt.rohloff@njit.edu
Homepage: https://web.njit.edu/~rohloff/
Description: Researchers in the NJIT cybersecurity research center have written and published one of the most advanced open-source libraries for post-quantum cryptography and homomorphic encryption. This library, called PALISADE, is being increasingly used by researchers and designers outside of NJIT. We have been taking pains to design the library using software engineering industry best practices. As the library and its user base grows, we seek to improve software quality through software testing using the Google C++ unit testing libraries. We are seeking motivated student with C++ experience who can help us implement unit tests using the Google C++ unit testing library for our open-source cryptography library.
Prerequsities: Experience with C++, such as in CS280
Blockchain Applications
Supervisor: Qiang Tang
Email: qiang.tang@njit.edu
Homepage: https://web.njit.edu/~qiang/
Description: Blockchain technology is currently one of the potentially most disruptive technologies that foster re-shaping the new Internet infrastructure and re-building trust in open networks. We would like to re-consider many traditional settings and rebuild them on top of the decentralized infrastructures.
1a. blockchain based anonymous e-commerce platform. Decentralized market place enabling digital commerce without relying on a central party like Amazon or Ebay, furthermore, we would make the transactions anonymous but the participants can be hold accountable if dispute happens.
1b. blockchain based financial applications: exploring decentralized and autonomous financial services, such as prediction market, or new payment system.
1c. blockchain based IoT system: study how to tailor and re-design blockchain protocols for lightweight and/or fast-response IoT applications.
Prerequisites: Programming proficiency, and ability to learn new technology quickly. Some knowledge about cryptography and bitcoin / blockchain technology is a plus.
NetExplorer
Supervisor: Jason Wang
Email: jason.t.wang@njit.edu
Homepage: http://web.njit.edu/~wangj
Description: The goal of NetExplorer is to develop a suite of algorithms, tools and web servers for inferring biological, social and transport networks using graph mining algorithms. Specifically, we design, develop, andimplement new software for (1) reconstructing networks using a data cleaning approach; (2) inferring networks using deep learning; (3) predicting missing links integrated, heterogeneous networks; and (4) reverse engineering networks using Big Data technologies such as Apache Spark and Hadoop in the cloud.
Prerequisites: The student is expected to collect and clean data. Depending on the student's background and expertise, the student will either implement his/her own algorithms or use existing tools to mine the data for network inference. Knowledge of data science languages such as Python, Java, R, Matlab, Hadoop or Spark is recommended, but not required.
Deep Learning for DNA Sequence Pattern Recognition
Email: zhi.wei@njit.edu
Homepage: http://web.njit.edu/~zhiwei
Description: In this project, we will apply deep neural networks to recognize interesting regulatory patterns in DNA sequence. For sequence data, classical machine learning methods cannot operate on the sequence directly, and thus need to pre-define features for model training. Features can be extracted from the sequence based on prior knowledge. As a results, the success of these conventional methods heavily replies on human engineered features. Deep neural networks can help circumventing the manual extraction of features by learning them from data. In addition, deep neural networks can capture nonlinear dependencies in the sequence and interaction effects and span wider sequence context at multiple genomic scales. In the past few years, attesting to their utility, deep neural networks have been successfully applied to quite a few applications of DNA sequence mining. In this project, we will investigate how deep neural networks work using simulated and real data. We will design several testing cases for finding out the connection between key deep learning parameters and their impact on prediction performance.
Prerequisites: Python or Java programming; knowledge of machine learning.
Quantitative Stock Selection and Trading
Supervisor: Zhi Wei
Email: zhi.wei@njit.edu
Homepage: http://web.njit.edu/~zhiwei
Description: In this project, we will develop and use quantitative approaches to select/trade stocks. We will identify a candidate list of potential factors that may assist in predicting stock returns (valuation, growth, sentiment information from ER, twitter, etc.). The techniques we may use include Machine Learning, Data mining, Nature language Processing, and Time Series Data Analytics. We will collect and clean historical price data, financial report data and social media data. We will develop, implement and back test quantitative trading strategies over historical data.
Prerequisites: Python or R programming; have taken some quantitative courses (algorithms, statistics, etc). Finance knowledge is a plus but not required.
Health Technologies and User Experience
Supervisor: Yvette Wohn
Email: wohn@njit.edu
Homepage: socialinteractionlab.com
Description: There are numerous technologies that can support healthy living, ranging from emotional support via social media to chronic disease management. Anyone can build or make a health technology, but it is essential to conduct user research to ensure that the technology is effective. Our lab is currently working on a number of different health technology projects in collaboration with researchers at Rutgers, University of Maryland, and Microsoft research among others. Current projects include: 1) Utilizing social media to improve mental health, 2) Using web tools to break sedentary behavior, 3) Identifying opportunities of telehealth for women's health, 4) Using games for stroke rehabilitation. You may join existing projects or propose your own area of interest.
Prerequisites: For front-end development, experience in Android and/or iOS programming is desired, for back-end development using Javascript and SQL is required. For UX research, no prior experience is required, but student must have strong interest in reading about health technologies and learning empirical methods of human subject research. If you have no prior experience, you must anticipate working in the lab for at least two semesters.
Understanding Livestreaming and eSports
Supervisor: Yvette Wohn
Email: wohn@njit.edu
Homepage: socialinteractionlab.com
Description: Livestreaming and esports are relatively new cultural trends that are recreational activities that require high technology specifications. Current projects are aimed at understanding more about people's behaviors in these environments with the aim of developing better systems, such as 1) Understanding virtual currency/ economies, 2) Relationships between streamers and viewers, 3) Cultural practices in esports. Based on your expertise and/or interest, you will be conducting qualitative or quantitative research in a collaborative environment with researchers from other universities.
Prerequisites: Strong interest in topic matter, voracious reader, good at interacting with people.
Write-and-Learn: Promoting Meaningful Learning Through Concept Map-Based Formative Feedback On Writing Assignments
Supervisor: Brook (Yi-Fang) Wu
Email: yi-fang.wu@njit.edu
Homepage: http://web.njit.edu/~wu
Description: The primary goal of meaningful learning is to deliver course content in innovative ways that allow students to learn and then apply. As a pedagogical strategy, Writing-to-Learn activities use writing to improve students; understanding of course content. We are developing an enhanced "Write-and-Learn" framework to generate automated formative feedback through comparing the concept maps constructed from teaching materials and students' writing assignments. Our work aims to (1) evaluate how effective the automated formative feedback is on the acquisition and development of conceptual knowledge, and (2) explore how such formative feedback can be utilized to scaffold and promote meaningful learning. We are developing Write-and-Learn system to generate automated formative feedback by taking advantage of the concept maps constructed from instructors' lecture notes and individual students' writing assignments to improve students' meaningful learning of conceptual knowledge in WTL activities. We are looking for students to participate in the design, development, maintenance, and evaluation of the research prototype, as well as the design and execution of the research studies in all facets of the project.
Prerequisites: Preferences will be given to students who (1) have solid skills and proven experience in building front-end web applications using HTML, CSS, and JavaScript; (2) have knowledge of working with server-side web application languages and frameworks (preference given to experience with Python and Django); (3) have some experience with relational database engines (e.g. MySQL, PostgreSQL,SQL Server); and (4) have a good understanding of general system administration and web security best practices.
Simulating Spatial Meme Diffusion
Supervisor: Xinyue Ye
Email: xinyue.ye@njit.edu
Description: A meme is an idea, behavior, or style that spreads from person to person. The dynamics of meme has been facilitated by the wide adoption of social media apps, along with the widespread of mobile devices. Social media messages contain a wealth of location information, which can help us understand some of the phenomena associated with the geographic location and reveal the law hidden behind the phenomenon. Funded by NSF, Spatial Social Network Simulator implements the methodological advances in an open source environment of Python for exploring spatial meme diffusion, using twitter data as the case study. Synthesis of spatial and social network analytics has contributed to our understanding of the spatiotemporal effect in the information diffusion process. This open source toolkit aims to explore the pattern of information diffusion over time on the social network, so it displays a complete process that starts with preparing a network and ends with demonstrating how information propagates over the network.
Building a Framework of Software and Hardware for Real-time Predictive Analytics on Social Networks
Supervisor: Andrew Sohn
Email: andrew.sohn@njit.edu
Description: Social networks are graphs that change constantly to reflect the current state of the mindsets and behaviors of users, organizations, institutions, and even countries. Capturing major changes is critical to the success of social networks, as analytics would allow them to predict what may come next and therefore be prepared with possible actions. Capturing major changes in real-time, however, is a challenge. This project builds a framework comprised of both hardware and software to address the problem of capturing critical changes in social networks. In particular, an initial cluster of 16 machines is currently being built to compute changes in real-time, while spectral graph partitioning software is being designed and implemented for analytics on the initial cluster. The outcome of the project is a prototype framework to demonstrate that dynamic spectral graph partitioning on a cluster of machines can enable real-time predictive analytics on large-scale social-networks. The qualifications of the principal investigator for this project include the contributions to the NASA Ames Research Center as well as Lawrence Berkeley National Laboratory on large-scale graph partitioning for computational science under a NASA University Joint Venture faculty fellowship. For the efforts, the PI was recognized as a NASA Education Research Pioneer.
Interactive Mix Reality Platform
Supervisor: Margarita Vinnikov
Email: margarita.vinnikov@njit.edu
Description: Students will have an opportunity to be part of developing interactive Mix Reality (XR) platform for serious games and industrial partners. Students will develop virtual and augmented reality and cross-modal visual, audio, and/or haptic user interfaces. They will device new techniques for improving navigation in XR (walking, driving, and flying and new approaches of the head and body tracking for XR interaction. Finally, students will help in developing and testing new gaze-contingent interfaces for the new generation of head-mounted displays.
Prerequisites: For students interested in my work would encourage to develop Unity 3D skills. Any other software that they would like to learn and experiment with for the purpose to create VR/AR/XR applications would also be supported.
Science and Engineering in Program Repair
Supervisor: Ali Mili
Email: mili@njit.edu
Description: For the past ten years, researchers in software engineering have been working on developing automated tools for program repair. We are interested in evolving theoretical foundations for this discipline, and to analyze the impact of these on the state of the art and the state of the practice in program repair.
Mining Big Data through Deep Learning
Supervisor: Jason Wang
Email: Jason.t.wang@njit.edu
Description: We are designing and implementing new deep learning algorithms for mining big data. We have developed a 3D-atrous convolutional neural network, used it as a deep visual feature extractor, and stacked convolutional long short-term memory networks on top of the feature extractor. This allows us to capture not only deep spatial information but also long-term temporal information in the data. In addition, we use stacked denoising autoencoders to learn latent representations of the data, to construct feature vectors suitable for classification. Currently, we are building a semi-supervised deep learning framework with generative adversarial networks (GANs) for event prediction. Such a framework is suited for big data that have few, incomplete, imperfect, missing, noisy or uncertain training data.
Prerequisites: Experience with Python, TensorFlow, Keras or PyTorch.
AI Deep Learning in Medical Image Analysis and Other Applications
Supervisor: Frank Shih
Email: shih@njit.edu
Homepage: http://web.njit.edu/~shih
Description: Artificial intelligence (AI) applied on medical image analysis for early detection, diagnosis, and treatment of diseases has been extremely important in human healthcare. Conventionally, meaningful features were primarily designed by human experts based on their knowledge of target domains. However, deep learning has relieved such obstacles by absorbing the feature engineering step into a learning step. Instead of extracting features in a hand-designed manner, deep learning requires only a set of data with minor preprocessing and then discovers the informative representations in a self-taught manner. My research group utilizes traditional AI techniques as well as the recently developed deep learning approaches for medical image analysis, ecological data classification, and other applications.
Prerequisites: The students are expected to have AI and image processing background. They can investigate existing algorithms or develop their own architectures to run on big data. Knowledge of data science languages, such as Python, Java, C, Matlab, is recommended, but not required.