Research in Computer Science

A stable, safe, and resilient cyberspace is vital for our economic and societal well-being. This concentration helps students learn how to fortify cyber networks, combat threats, and foster “white hat” hacking. Researching systems allows students to improve real-world systems to make them stronger and more secure. This also includes data-driven analysis of privacy and social networks. After graduation, our students often work in private corporations or governments.

Labs: OSIRIS, CCS

Sample research projects:

Damon McCoy, one of the department's faculty members, researches Cybersecurity for Democracy. Cybersecurity for Democracy is a multi-university center for problem-driven research and research-driven policy. They conduct cutting-edge cybersecurity research to understand better the distorting effects of algorithms and AI tools on large online networks and work with platforms and regulators to help all parties understand the implications of our findings and develop solutions.

Learn more about Cybersecurity for Democracy. 

Justin Cappos works on software supply chain security systems which are used in practice.  This includes the TUF project, which is used widely in the cloud and other domains, Uptane which is used in automobiles and IoT devices, in-toto which is the attestation standard used widely across industry, and similar projects.
Learn more about Professor Cappos's Work

The organization and governance of large volumes of data. This concentration allows for retaining data obtained from a large number of sources — from a large city, to individuals, and anywhere in between — and ensures a high level of data quality for analytical purposes. The visualization of such data elegantly brings structure and simplicity to it.

Labs: CUSP

Sample research projects:

Associate Professor Rumi Chunara works at the intersection of Big Data and Public Health, developing models using data from in the hospital as well as outside the hospital, representing social and environmental factors.

For example, satellite imagery and deep learning are combined to identify and quantify the types of greenspaces across an entire urban area. Similarly we have used other data such as social media to understand social factors across places, and clinical notes to assess how large language models work across different hospitals in New York City.

Learn more about Chunara's study

For students who are interested in learning game programming and taking part in game development and design. Computer graphics, human-computer interaction, artificial intelligence, and allied computational fields all play a role in this burgeoning industry. Art and engineering intersect to create innovative game environments that captivate players.

The theoretical study of computer science allows us to better understand the capabilities and the limitations of exactly what problems computers can solve, and when they can solve those problems efficiently. New theory helps pave the way for algorithmic breakthroughs that engineers can build on to create new solutions and technology. At NYU Tandon, the Algorithms and Foundations group is composed of researchers interested in applying mathematical and theoretical tools to a variety of disciplines in computer science, from machine learning, to computational science, to geometry, to computational biology, and beyond.

Sample research projects:

Lisa Hellerstein is a co-author of "How to Quickly Determine Who Won an Election."  

This paper gives approximation algorithms for a stochastic sequential ordering problem where d candidates compete in an election with n voters.  The problem is to determine the order in which to inspect the votes cast by the voters, to minimize the expected time to determine the election winner.  

This paper will appear in the Innovations in Theoretical Computer Science (ITCS) conference in Jan. 2024. 

Yi-Jen Chiang, Christopher Musco and student Mengxi Wu have developed a novel greedy, online streaming algorithm for the key time steps selection problem. Current computer simulations often generate time-varying volume data that exceeds both the available storage capacity and bandwidth for transferring simulation output to storage. Accordingly, it becomes necessary to select (and store) key time steps on the fly. However, solving the key time step selection problem in the in situ setting is very challenging, as we can only process data in one pass in an online streaming fashion, using a small amount of main memory and fast computation. Previously no solution with a theoretical guarantee is known. In this paper, we develop the first streaming approach for in situ selection of key time steps from time-varying volume data, with strong theoretical guarantees on the approximation quality and number of segments stored. It also works well in practice. Experiments demonstrate the efficacy of our new techniques. Read the Paper (including talk slides and code)

Christopher Musco and Algorithms and Foundations Ph.D. student Majid Daliri wrote a paper collaborating with researchers from NYU Tandon's VIDA lab on randomized sketching algorithms for inner product estimation. This problem is essential in applications across database systems, like join size estimation, as well as in dataset search and augmentation applications. The authors' work gives the first asymptotic improvement on the accuracy of sketching methods based on the famed Johnson-Lindenstrauss lemma, which are widely used in practice. Read the paper or take a look at their recent follow-up work!