Amil Dravid Named a Finalist in the CRA Undergraduate Researcher Awards

Dravid made significant contributions to research in computer vision

Amil DravidNorthwestern Engineering’s Amil Dravid was named a finalist in the Computer Research Association (CRA) 2023 Outstanding Undergraduate Researcher Awards. The nationwide award program, sponsored in 2023 by Microsoft Research, recognizes undergraduate students who demonstrate exemplary potential in computing research.

Dravid is a fourth-year student in computer science at the McCormick School of Engineering advised by Aggelos Katsaggelos, Joseph Cummings Professor of Electrical and Computer Engineering and (by courtesy) of computer science and director of the Image and Video Processing Laboratory. Dravid is also pursuing a minor in statistics at Northwestern’s Weinberg College of Arts and Sciences.

Dravid’s research interests broadly center on explainability and reliability in computer vision, a subfield of artificial intelligence (AI) that trains models to process, interpret, and analyze inputs like images and videos. By leveraging an approach that disentangles, or separates, computer vision concepts and high-level abstractions, he develops tools to explain the internal mechanisms of machine learning algorithms and correct learned biases.

In 2022, Dravid completed an internship with the Microsoft Research Computer Vision Group, a Caltech Summer Undergraduate Research Fellowship with Pietro Perona; and a visiting researcher position at the Berkeley Artificial Intelligence Research Lab with Alexei Efros. He has published four papers as first author and contributed to an additional three in venues including the Association for the Advancement of Artificial Intelligence, IEEE International Conference on Acoustics, Speech, & Signal Processing, the academic journal Radiology, and the IEEE/CVF Computer Vision and Pattern Recognition Conference.

“It is an honor to be recognized as a finalist for the CRA Outstanding Undergraduate Researcher Award,” Dravid said. “I am humbled to be among a cohort of such amazing young researchers. I am very grateful for my mentors, specifically Professor Aggelos Katsaggelos, who has advised me since my freshman year.”

Dravid is working on a project that enables convolutional neural networks (CNNs) to generalize to new environments for image-based tasks. His novel approach combines features learned by a CNN with hand-engineered features that encode properties into invariant representations. His method outperforms the state-of-the-art for classification accuracy in different environments, a task known as domain adaptation or generalization.

Keen to develop trustworthy AI systems with the potential to improve society and quality of life, Dravid is also interested in how a class of networks called generative adversarial networks (GANs) can be applied in the healthcare setting. GANs mimic an existing distribution of data to generate new images.

Aiming to understand why CNNs make certain diagnoses on medical images, Dravid and his team developed a generative adversarial network (GAN) framework to visually explain what a medical image-based classifier focuses on in its binary predictions. The method — called medXGAN —produces fine-grained visualizations that outperform popular baselines in disease feature localization and explanatory ability. Dravid also proposed new evaluation methods to measure how strong detected features correlate with a classifier’s decision and the degree to which to the features are localized.

The project “Visual Explanations for Medical Classifiers through a Generative Latent Space,” was performed in collaboration with Katsaggelos; Oliver Cossairt, associate professor of computer science and of electrical and computer engineering; Florian Schiffers, a PhD student in computer science; and Boqing Gong, a research scientist at Google.

Dravid mentors two second-year undergraduate students in computer science research, guiding them in developing projects and grant proposals for submission to Northwestern’s Office of Undergraduate Research.

In 2021, Dravid received an honorable mention in the CRA Outstanding Research Award Program. He was also awarded the 2021 Barry Goldwater Scholarship, an honor that funds students who intend to pursue research careers in the natural sciences, mathematics and engineering. Beyond his academic achievements, Dravid is a nationally ranked Taekwondo competitor.

Looking ahead, Dravid aims to pursue a PhD in computer vision and machine learning.

“I aspire to become a professor, mentoring the next generation of thinkers and conducting cutting-edge research,” Dravid said.

 

McCormick News Article