Arnab Neelim Mazumder

Logo

Postdoc@LANL, Ex-PhD@UMBC, Ex-Visiting Research Scholar@JHU, Ex-Intern@Infineon Technologies

View My GitHub Profile

Postdoc, AI/ML Researcher

Resume, CV, Google Scholar

About

I am a postdoc at Los Alamos National Laboratory with a PhD in Computer Engineering from the University of Maryland Baltimore County and with both academic and internship experience in developing explainable deep learning modules for audios and images as well as their deployment on embedded devices. Specialties include but are not limited to Python, MATLAB, C Programming, TensorFlow, PyTorch, and Verilog. Previously, I worked as an intern at Infineon Technologies with a focus on building transparent and interpretable AI models. Current research interests focus on hardware-aware fine-tuning of DNNs, explainable AI, compression techniques for Embedded AI, and AI/ML for vision, audio, time-series, environmental and healthcare applications.

Education

Work Experience

Postdoctoral Research Associate@ Los Alamos National Laboratory (November 2024 - Present)

Visiting Research Scholar @ Johns Hopkins University (September 2023 - October 2024)

Graduate Research Assistant @ University of Maryland Baltimore County (May 2022 - August 2024)

Intern - Machine Learning @ Infineon Technologies (June 2022 - December 2023)

Technical Skills

Projects

Reg-Tune: Regression-Based Fine-Tuning Approach for DNNs on Embedded Platforms

Worked on the Google Speech Commands and CIFAR-10 dataset to develop low energy-consuming models on FPGA and NVIDIA Jetson Nano processors that provide near-optimal (100x reduction in model size)} results for multiple objectives using Tensorflow, PyTorch, QKeras, Librosa, MATLAB, TensorRT and Vivado HDL.

Reg-Tune

XAI-Increment: Explainable-AI-focused Incremental Learning

Designed the XAI-inspired incremental learning framework (at least 1% accuracy improvement throughout all incremental sessions)

XAI

Low-power Multi-modal CNN for Respiratory Symptom Detection

Built and deployed the optimized multi-modal framework for respiratory symptoms detection with an End-to-end CNN architecture (1.5x improvement for energy efficiency with no accuracy degradation) using Tensorflow, PyTorch, Keras, and Vivado HDL.

Multi-modal

Low-power Digital Accelerators for Human Activity Recognition

Built and deployed the optimized DNN frameworks for human activity recognition on digital accelerators (5.7x improvement for energy efficiency) using Tensorflow, PyTorch, Keras, and Vivado HDL.

HAR-model

All Publications

  1. Utteja Kallakuri, Bharat Prakash, Arnab Neelim Mazumder, et al., 2024. “ATLAS: Adaptive Landmark Acquisition using LLM-Guided Navigation” First Vision and Language for Autonomous Driving and Robotics (VLADR) Workshop, (CVPR).
  2. Arnab Neelim Mazumder and T. Mohsenin, “Reg-TuneV2: A Hardware-Aware and Multiobjective Regression-Based Fine-Tuning Approach for Deep Neural Networks on Embedded Platforms,” in IEEE Micro 2023.
  3. Arnab Neelim Mazumder, Farshad Safavi, Maryam Rahnemoonfar, and Tinoosh Mohsenin. “Reg-Tune: A Regression-Focused Fine-Tuning Approach for Profiling Low Energy Consumption and Latency,” ACM Transactions of Embedded Computing and Systems (TECS) 2023.
  4. Arnab Neelim Mazumder, Niall Lyons, Ashutosh Pandey, Avik Santra and Tinoosh Mohsenin, “Harnessing the Power of Explanations for Incremental Training: A LIME-Based Approach,” 2023 31st European Signal Processing Conference (EUSIPCO), Helsinki, Finland.
  5. Hasib-Al Rashid, Haoran Ren, Arnab Neelim Mazumder, Mohammad M. Sajadi & Tinoosh Mohsenin, “A Re-configurable Software-Hardware CNN Framework for Automatic Detection of Respiratory Symptoms,” The Science behind the COVID Pandemic and Healthcare Technology Solutions 2022
  6. Dennis V. Christensen, Regina Dittmann, et al., “2022 roadmap on neuromorphic computing and engineering,” IOP Publishing Limited.
  7. Arnab Neelim Mazumder, and Tinoosh Mohsenin. “A fast network exploration strategy to profile low energy consumption for keyword spotting.” tinyML Research Symposium 2022.
  8. S. Rafatirad, H. Homayoun, Z. Chen, S. M. Pudukotai Dinakarrao, et al., “Sensornet: An educational neural network framework for low-power multimodal data classification,” in Machine Learning for Computer Scientists and Data Analysts 2022.
  9. Arnab Neelim Mazumder et al., “A Survey on the Optimization of Neural Network Accelerators for Micro-AI On-Device Inference,” in IEEE Journal on Emerging and Selected Topics in Circuits and Systems (JETCAS) 2021.
  10. Morteza Hosseini, Nitheesh Kumar Manjunath, Bharat Prakash, Arnab Neelim Mazumder, Vandana Chandrareddy, Houman Homayoun, and Tinoosh Mohsenin, “Cyclic sparsely connected architectures for compact deep convolutional neural networks,” in IEEE Transactions on Very Large Scale Integration (TVLSI) Systems 2021.
  11. Aidin Shiri, Bharat Prakash, Arnab Neelim Mazumder, Nicholas R Waytowich, Tim Oates, and Tinoosh Mohsenin, “An energy-efficient hardware accelerator for hierarchical deep reinforcement learning,” in IEEE 3rd International Conference on Artificial Intelligence Circuits and Systems (AICAS) 2021.
  12. Hasib-Al Rashid, Arnab Neelim Mazumder, Utteja Panchakshara Kallakuri Niyogi, and Tinoosh Mohsenin, “CoughNet: A flexible low power CNN-LSTM processor for cough sound detection,” in IEEE 3rd International Conference on Artificial Intelligence Circuits and Systems (AICAS) 2021.
  13. Arnab Neelim Mazumder et al., “Automatic Detection of Respiratory Symptoms Using a Low-Power Multi-Input CNN Processor,” in IEEE Design & Test 2021.
  14. Aidin Shiri, Arnab Neelim Mazumder, Bharat Prakash, Houman Homayoun, Nicholas R Waytowich, and Tinoosh Mohsenin, “A hardware accelerator for language-guided reinforcement learning,” in IEEE Design & Test 2021.
  15. Morteza Hosseini, Mohammad Ebrahimabadi, Arnab Neelim Mazumder, et al. “A fast method to fine-tune neural networks for the least energy consumption on FPGAs,” in Hardware Aware Efficient Training (HAET) workshop of International Conference on Learning Representations (ICLR) 2021
  16. Morteza Hosseini, Haoran Ren, Hasib-Al Rashid, Arnab Neelim Mazumder, et al., “Neural networks for pulmonary disease diagnosis using auditory and demographic information,” in epiDAMIK 2020: 3rd epiDAMIK ACM SIGKDD International Workshop on Epidemiology meets Data Mining and Knowledge Discovery, 2020.
  17. Arnab Neelim Mazumder, Hasib-Al Rashid, et al., “An Energy-Efficient Low Power LSTM Processor for Human Activity Monitoring,” in IEEE 33rd International System-on-Chip Conference (SOCC) 2020.
  18. Haoran Ren, Arnab Neelim Mazumder, Hasib-Al Rashid, Vandana Chandrareddy, Aidin Shiri, Nitheesh Kumar Manjunath, and Tinoosh Mohsenin, “End-to-end scalable and low power multi-modal CNN for respiratory-related symptoms detection,” in IEEE 33rd International System-on-Chip Conference (SOCC) 2020.
  19. Aidin Shiri, Arnab Neelim Mazumder, Bharat Prakash, Nitheesh Kumar Manjunath, Houman Homayoun, Avesta Sasan, Nicholas R Waytowich, and Tinoosh Mohsenin, “Energy-efficient hardware for language guided reinforcement learning,” in Proceedings of the Great Lakes Symposium on VLSI (GLSVLSI) 2020.
  20. Arnab Neelim Mazumder, Emon Dey, and Sharmin Majumder, “A Color Image Watermarking Scheme Employing the Features of Directive Contrast in the DWT-SVD Domain,” in IEEE 5th International Conference for Convergence in Technology (I2CT) 2019.
  21. Emon Dey, Sharmin Majumder, Arnab Neelim Mazumder, “A new approach to color image watermarking based on joint DWT-SVD domain in YIQ color space,” in 3rd International Conference on Electrical Information and Communication Technology (EICT) 2017.

Patents

  1. Niall Lyons, Arnab Neelim Mazumder, Avik Santra, Anand Dubey, Ashutosh Pandey, “Managing data drift in machine learning models using incremental learning and explainability,” US Patent App. 18/178,351.
  2. Niall Lyons, Arnab Neelim Mazumder, Avik Santra, Anand Dubey, Ashutosh Pandey, “Building generalized machine learning models from machine learning model explanations,” US Patent App. 18/178,223.

Talks and Presentations

Activities and Peer Review Services

Contact