Workshop Details

Nokia Bell Labs will host the Spring 2023 DIMACS Mixer featuring an invited presentation by Professor Arian Maleki of Columbia University. The event will also include a series of five-minute talks and a reception to help people get to know each other and learn more about the breadth of research being conducted within our community.

The event will take place in the Oak Room and is open to everyone. We especially welcome students and postdocs.

If you plan to attend, please email Lazaros Gallos This email address is being protected from spambots. You need JavaScript enabled to view it., with your name, affiliation, and citizenship. This information is needed to enter Nokia Bell Labs. You can still join the event if you have not sent this information, but you will need to register at the main entrance on the day of the event.

Schedule:

1:45 – 1:55 – Attendees arrive at Bell Labs and are directed to the Oak Room

2:00 – 2:05 – Welcome, Iraj Saniee, Nokia Bell Labs

2:05 – 2:15 – A Short Introduction to DIMACS, David Pennock, DIMACS

2:15 – 3:15 – Some Recent Mathematical Advances in Imaging Sciences, Arian Maleki, Columbia

3:15 – 3:20 – Introduction to the Research Overview Sessions, Lisa Zhang, Nokia Bell Labs

3:20 – 4:00 – 5-minute research talks by Nokia Bell Labs and DIMACS members

• Learning to Communicate - Examples of Acoustic Echo Cancellation and Digital Pre-Distortion, Carl Nuzman, Math & Algorithms Group, Nokia Bell Labs
• Statistical Inference from Privacy-Protected Data, Ruobin Gong, Dept. of Statistics, Rutgers University
• Analyzing Neural Network Performance For Value-Based Deep Reinforcement Learning, Atefeh Mohajeri, Modeling & Optimization Group, Nokia Bell Labs
• Efficient Characterization of Robot Controller Dynamics with Confidence Guarantees using Limited Data, Ewerton Rocha Vieira, DIMACS
• TRAIL – Trouble-ticket Routing with AI Large Language Models, Mohamed Trabelsi, Stats & Data Science Group, Nokia Bell Labs
• Streaming Algorithms for Problems on Massive Graphs, Prantar Ghosh, DIMACS

4:00 – 5:00 – Reception

Featured Presentation: Some Recent Mathematical Advances in Imaging Sciences

Abstract: In the last decade, we have witnessed major progress in our mathematical understanding of imaging systems, such as astrophotography, x-ray crystallography, holography, and synthetic aperture radar. Despite this progress, the mainstream theoretical frameworks have fallen short of answering many fundamental questions that appear in practice. For instance, for solving the inverse problem corresponding to each type of imaging system, one has access to a wide range of numerical algorithms, each with an "optimality result" associated with it. This leaves a potential user bewildered on the choice of the algorithm. While these algorithms offer "optimal" performance from different theoretical viewpoints, they show very different performances in practice. Moreover, in many imaging systems, such as MRI, the user can tweak the measurement kernel to achieve a higher resolution, but the optimal choice of such measurement kernels is unknown. Answering such questions can provide new opportunities for creating more efficient imaging systems.

In this talk, we will discuss a theoretical framework that aims to address such questions. This framework is inspired by asymptotic analysis, popular in Statistics and Statistical Physics, which accurately characterizes the performance of different algorithms in the regimes where the number of sensors and the number of pixels in the recovered image are large. We show how asymptotic analysis addresses the above questions. We will also review some recent advances in this field and present some of the open problems.

Speaker Bio: Arian Maleki is an associate professor in the Department of Statistics at Columbia University. Arian received his PhD from Stanford University in 2011. Before joining Columbia University in 2013, he was a postdoctoral scholar at Rice University. Arian's research interests span a range of topics in high dimensional statistics, mathematical analysis of imaging systems, and machine learning.