Biography
Aritra Dutta is an Assistant Professor at the AI Initiative, University of Central Florida (UCF). He is primarily affiliated with the Department of Mathematics with secondary joint affiliation with the Department of Computer Science. Previously, he was an Assistant Professor at the Department of Mathematics and Computer Science, University of Southern Denmark (SDU). Additionally, Dr. Dutta is affiliated with the Pioneer Centre for AI (P1), Denmark. Dr. Dutta’s research interests include (stochastic) optimization, distributed computing (with application in federated learning), numerical linear algebra, machine learning, low-rank approximation, and its applications in computer vision (image and video analysis). Additionally, he is interested in different video understanding tasks (object detection and tracking, action recognition, etc.) in computer vision.
Potential Ph.D. Students & Postdocs
I am always interested in hearing from potential students wanting to pursue a Ph.D. If you want to work with me, here are a few notes to help you with the application process.
Please do not contact me before you have read all the relevant parts of this page. I will know if you have not, and I will ignore your message.
If you read the rest of this page, I assume it’s because you want to work with me; otherwise, please follow the instructions on the general graduate admissions at the College of Graduate Studies website at UCF.
Other useful links are: Applying to UCF.
Jump to the following sections:
Ph.D. Students PostdocsPh.D. Students
I am looking forward to hiring three to four PhD students. Two of these positions can start as early as the Summer 2024 semester. The remaining positions can begin in Fall 2024 and Spring 2025. For Fall (and Summer) 2024, the application deadlines and requirements are found here. For CS and Math Ph.D. admission in Fall 2024, the international application deadline is on 1st December 2023; see here.
All admitted students to work with me at UCF, can have flexible choice of the department; it can be either mathematics or computer science, based on the candidate’s competence. An ideal candidate would have strong foundational knowledge in either mathematics (optimization, linear algebra, numerical analysis) or computer science (machine learning, distributed computing, and/or computer vision). An ideal candidate in mathematics is expected to pick up prototyping skills quickly to execute proof-of-concept. Similarly, a CS candidate is expected to understand the behind-the-scenes mathematical heuristics to implement better. If a candidate has publication(s) in top-tier conferences (CS-ranking list) or Q1 mathematics journals, then they would earn extra brownie points.
The positions come with attractive assistantship, tuition support, the chance to work and collaborate with world leaders in their fields (inside UCF and outside), and an opportunity to be associated with the UCF AI initiative (UCF Aii), and UCF Center for Research in Computer Vision (UCF CRCV). Interested candidates, please refer to the following for more details.
Selected PhD students receive the benefits of the competitive research assistantship, which supports students for the duration of their graduate studies. The assistantships for graduate students at the UCF Aii are even more attractive. Therefore, funding is not an issue: only admission is. I know what you are going to do next; therefore, …
Please do not send me attachments (especially resumes) or ask me questions about your chances of admission. Our decisions are made by a committee that considers the applicant pool. Furthermore, your decision hinges heavily on material, such as recommendation letters, that you cannot send me. Therefore, I cannot give you any objective evaluation of your chances.
You can help your chances in two ways. First, you must have a genuine interest in my work. Second, do not be clueless. If, for instance, you’re asked to not do certain things (see above) but you do so anyway, it calls into question your reading comprehension and makes me much less interested in you.
To demonstrate interest in my work, you must show that (1) you have some idea of what I do, (2) you have some background that at least intrigues me, and (3) some of my work interests you. The best way to do this is to show that you’ve read some of my papers (and possibly build upon them). Write me a letter so that it:
- indicates that you have read this page by including the phrase Disney,
- summarizes your preparation (academics, work experience),
- describes highlights: particularly tough courses you took, especially profound texts you read, etc.,
- explains what you know of my work (which papers you have read),
- explains why you want to do a PhD with me.
We will pick up a serious communication from there.
I realize I am asking for a lot: you may have a form letter and attachment you have been sending out, and it probably does not fit the criteria above. But remember that you are asking for a lot, too: you are implicitly asking me to commit several years of time, funding, and emotional energy. Do not you think that kind of commitment deserves more than a form letter?
Postdocs
I am interested in postdocs. I will work to accommodate outstanding candidates.
If you are close to completing a PhD or already have one, you should know how to establish communication with a scientific colleague, so I will not belabor the point. Nevertheless, you would do well to glance quickly at the instructions for PhD applicants, above.
Marco Canini from KAUST inspired me to create this page.
Return to TopSelected Publications
- Personalized Federated Learning with Communication Compression, El Houcine Bergou, Konstantin Burlachenko, Aritra Dutta, and Peter Richtárik, Transactions on Machine Learning Research, November 2023.
- On the Convergence Analysis of Asynchronous SGD for Solving Consistent Linear Systems, Atal N. Sahu, Aritra Dutta, Aashutosh Tiwari, and Peter Richtárik, Linear Algebra and its Applications—Elsevier, Vol. 663, pp. 1–31, 2023.
- Rethinking Gradient Sparsification as Total Error Minimization, Atal N. Sahu, Aritra Dutta, Ahmed M. Abdelmoniem, Trambak Banerjee, Marco Canini, and Panos Kalnis, in Thirty-fifth Conference on Neural Information Processing Systems, Vol. 34, pp. 8133–8146, 2021. (Designated for Spotlight—less than 3% out of 9,122 submissions)
- DeepReduce:A Sparse-tensor Communication Framework for Federated Deep Learning, Hang Xu, Kelly Kostopoulou, Aritra Dutta, Xin Li, Alexandros Ntoulas, and Panos Kalnis, in Thirty-fifth Conference on Neural Information Processing Systems, Vol. 34, pp. 21150–21163, 2021. (26% acceptance rate out of 9,122 submissions)
- GRACE: A Compressed Communication Framework for Distributed Machine Learning, Hang Xu, Chen-Yu Ho, Ahmed M. Abdelmoniem, Aritra Dutta, El Houcine Bergou, Konstantinos Karatsenidis, Marco Canini, and Panos Kalnis, 41st IEEE International Conference on Distributed Computing Systems, pp. 561–572, 2021. (19.8% acceptance rate)
- Best Pair Formulation & Accelerated Scheme for Non-convex Principal Component Pursuit, Aritra Dutta, Filip Hanzely, Jingwei Liang, and Peter Richtárik, IEEE Transactions on Signal Processing, Vol. 68, pp. 6128–6141, 2020.
- On the Discrepancy between the Theoretical Analysis and Practical Implementations of Compressed Communication for Distributed Deep Learning, Aritra Dutta, El Houcine Bergou, Ahmed M. Abdelmoniem, Chen-Yu Ho, Atal Narayan Sahu, Marco Canini, and Panos Kalnis, in Thirty-Fourth AAAI Conference on Artificial Intelligence, 34 (4), pp. 3817–3824, 2020. (20.6% acceptance rate out of 8,800 submissions)
- A Nonconvex Projection Method for Robust PCA, Aritra Dutta, Filip Hanzely, and Peter Richtárik, in Thirty-Third AAAI Conference on Artificial Intelligence, 33 (01), pp. 1468–1476, 2019. (Designated for oral presentation—less than 8.4% out of 7,700 submissions)
- Online and Batch Supervised Background Estimation via L1 Regression, Aritra Dutta and Peter Richtárik, WACV 2019–IEEE Winter Conference on the Applications of Computer Vision, pp. 541–550, 2019. (34% acceptance rate)
- Fast Detection of Compressively-Sensed IR Targets Using Stochastically Trained Least Squares and Compressed Quadratic Correlation Filter, Brian Millikan, Aritra Dutta, Qiyu Sun, and Hassan Foroosh, IEEE Transactions on Aerospace and Electronic Systems, Vol. 53, Issue 5, pp. 2449–2461, 2017.
- On a Problem of Weighted Low Rank Approximation of Matrices, Aritra Dutta and Xin Li, SIAM Journal on Matrix Analysis and Applications, Vol. 38, No. 2, pp. 530–553, 2017.
Classes Teaching Fall 2023
Matrices and Linear Algebra (MAS 3105)
