Huang Receives NSF and J.P. Morgan Research Awards

Fri May 10, 2019

Furong Huang, an assistant professor of computer science with an appointment in the University of Maryland Institute for Advanced Computer Studies (UMIACS), has received two separate research awards from the National Science Foundation and J.P Morgan to study neural networks and trading behavior over financial data streams, respectively.

The two-year NSF award for approximately $175,000 is part of the NSF’s Robust Intelligence (RI) program. The funding also falls under the NSF CISE Research Initiation Initiative (CRII), given to talented young faculty who are in their first two years of a tenure-track academic position.

Huang’s research will aim to design compressed neural network algorithms that are guaranteed to provide an optimal solution, generalize to unseen scenarios, and be robust to adversarial changes even if an attacker is given full knowledge of the algorithm.

Deep neural networks have elicited breakthrough successes in machine learning by achieving impressive accuracies on diverse tasks such as facial recognition, object identification, anomaly detection and monitoring assistance on a large scale. However, deep neural networks are not theoretically guaranteed to always perform well, and there is always a slight chance that they could fail in the presence of previously unseen data, or after small, imperceptible changes to the data.

For example, a home security system using a deep neural network facial recognition algorithm could mistake a stranger wearing pixelated sunglasses for the homeowner; a slight change of the environment, such as a rainy day, could cause a computer vision based autonomous driving vehicle to wrongly recognize a "STOP" sign as an outdoor commercial sign. The existence of such failed cases in widely used machine learning systems could put our daily lives—and even national security—at risk.

The methods developed via this research will provide theoretical bases that explain “black-box” deep neural networks and provide guarantees over their performance when applied to high-stakes problems.

The project will be integrated with graduate and undergraduate education, fostering collaboration between researchers from computer science, applied math, physics and business. Software programs developed via this project will be released as an open-source toolkit.

The J.P. Morgan Faculty Research Award, which comes with $150,000 in funding, is for the project, “Methods to Identify Communities and Trading Behavior over Financial Data Streams.”

As part of its first AI Research Awards, J.P. Morgan has presented 47 financial grants to university faculty and doctoral students for artificial intelligence research. The winners will study the use of AI and machine learning in areas including investment advice, risk management, digital assistants and trading behavior.

Huang’s project focuses on scalability of analysis to multiple data streams, and to fuse the data streams so that they represent communities of people and their interactions across financial products.

She is collaborating with Louiqa Raschid and Alberto Rossi, a professor and associate professor, respectively, in the Robert H. Smith School of Business. Raschid also holds an appointment in UMIACS.

The research aims to address a number of challenges, including customizing and extending machine learning approaches to fuse longitudinal time-series datasets, organized around people and financial products, and to analyze them at scale; proposing simple templates to identify community patterns; and connecting communities and their behavioral patterns to questions of interest to financial researchers.


Principled Methods for Learning and Understanding of Neural Networks” is supported by NSF grant #1850220 from the NSF’s Division of Information & Intelligent Systems.

PI: Furong Huang, assistant professor of computer science with an appointment UMIACS.

—Story by Melissa Brachfeld