Rahul Duggal

I'm an Applied Scientist with the AWS-AI Computer Vision group. Previously, I got my PhD in Computer Science from the Georgia Institute of Technology where I was advised by Polo Chau and Jimeng Sun.

My research focuses on empowering AI on edge devices (e.g. smartphones, smart watches). This requires developing neural networks that are:
  • Efficient to run on resource constrained platforms.
  • Robust to distributional changes in the test data.
  • Resilient to class imbalance in the training data.

If you work in related areas or are just curious about my work, don't hesitate to reach out to me via email.

Github  /  Google Scholar  /  Email  /  CV

Robust Efficient Edge AI: New Principles and Frameworks for Empowering Artificial Intelligence on Edge Devices

Rahul Duggal

PhD Dissertation, Georgia Tech, 2022

[Thesis][Slides][Video (coming soon)]

Towards Regression-Free Neural Networks for Diverse Compute Platforms

Rahul Duggal, Hao Zhou, Jun Fang, Shuo Yang, Yuanjun Xiong, Wei Xia

European Conference of Computer Vision (ECCV), 2022, Tel Aviv, Israel

[Paper] [Poster] [Video]

label neural architecture search, representation learning

MalNet: A Large-Scale Cybersecurity Image Database of Malicious Software

Scott Freitas, Rahul Duggal, Polo Chau

International Conference on Information and Knowledge Management (CIKM), 2022, Atlanta, Georgia

[Paper] [Demo] [Code]

label malware-classification, class-imbalance

Compatibility-aware Heterogeneous Visual Search

Rahul Duggal, Hao Zhou, Shuo Yang, Yuanjun Xiong, Wei Xia, Zhuowen Tu, Stefano Soatto

Computer Vision and Pattern Recognition (CVPR), 2021, Nashville, USA

[Paper] [Supplementary] [Poster] [Video]

label neural architecture search, visual search, compatible representations

HAR: Hardness Aware Reweighting for Imbalanced Datasets

Rahul Duggal, Scott Freitas, Sunny Dhamnani, Polo Chau, Jimeng Sun

IEEE Conference on Big Data (BigData), 2021, Orlando, USA

[Paper] [Slides] [Video]

label class-imbalance, early-exiting

CUP: Cluster Pruning for Efficient Deep Neural Networks

Rahul Duggal, Cao Xiao, Richard Vuduc, Polo Chau, Jimeng Sun

IEEE Conference on Big Data (BigData), 2021, Orlando, USA

[Paper] [Code] [Slides] [Video]

label filter-pruning, clustering, neural network compression

NeuroCartography: Scalable Automatic Visual Summarization of Concepts in Deep Neural Networks

Haekyu Park, Nilaksh Das, Rahul Duggal, Austin P. Wright, Omar Shaikh, Fred Hohman, Duen Horng Chau

IEEE Transactions on Visualization and Computer Graphics (TVCG), 2021

[Paper] [Demo]

label concept-visualization, neural networks

REST : Robust and Efficient Neural Networks for Sleep Staging in the Wild

Rahul Duggal*, Scott Freitas*, Cao Xiao, Polo Chau, Jimeng Sun (* = equal contribution)

The Web Conference (WWW), 2020, Taipei (Oral)

[Paper] [Arxiv] [Code] [Video]

label filter-pruning, noise-robustness, adversarial-training,

GCTI-SN: Geometry-inspired chemical and tissue invariant stain normalization of microscopic medical images

Anubha Gupta, Rahul Duggal, Shiv Gehlot, Ritu Gupta, Anvit Mangal, Lalit Kumar, Nisarg Thakkar, Devprakash Satpathy

Medical Image Analysis, 2020

[Paper] [Code]

label stain-normalization, singular value decomposition, basis alignment

PCSeg: Color model driven probabilistic multiphase level set based tool for plasma cell segmentation in multiple myeloma

Anubha Gupta, Pramit Mallick, Ojaswa Sharma, Ritu Gupta, Rahul Duggal

PLOS ONE, 2018


label stain-segmentation, level-sets

SD-Layer: Stain Deconvolutional layer for CNNs in Medical Microscopic Imaging

Rahul Duggal, Anubha Gupta, Ritu Gupta, Pramit Mallick

Medical Image Computing and Computer Assisted Intervention (MICCAI), 2017, Quebec City, Canada

[Paper] [Code] [Poster]

label tissue-classification, conv nets, optical-density colorspace

P-TELU : Parametric Tan Hyperbolic Linear Unit Activation for Deep Neural Networks

Rahul Duggal, Anubha Gupta

International Conference of Computer Vision Workshops (ICCVW), 2017, Venice, Italy

[Paper] [Poster]

label activation functions, deep learning

Overlapping Cell Nuclei Segmentation in Microscopic Images Using Deep Belief Networks

Rahul Duggal, Anubha Gupta, Ritu Gupta, Manya Wadhwa, Chirag Ahuja

Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP), 2016, India

[Paper] [Code] [Poster]

label stain-segmentation, deep belief networks

Pre-prints / Working Papers
IMBNAS: Neural Architecture Search on Imabalanced Datasets

Rahul Duggal, Shengyun (Anthony) Peng, Hao Zhou, Polo Chau

In Submission, 2022


labelneural architecture search, class-imbalance

ConceptEvo: Interpreting Concept Evolution in Deep Learning Training

Haekyu Park, Seongmin Lee, Ben Hoover, Austin Wright, Omar Shaikh, Rahul Duggal, Nilaksh Das, Judy Hoffman, Polo Chau

In Submission, 2022



Thanks to John Barron for the website design.