RadMamba: Efficient Human Activity Recognition through Radar-based Micro-Doppler-Oriented Mamba State-Space Model
in IEEE Transactions on Radar Systems (T-RS), vol. 4, pp. 261-272, 2026
Small devices. Useful intelligence.
Speech & intelligent sensing
We co-design sensing algorithms and computing hardware for real-time edge applications. Research includes speech denoising and synthesis, low-power keyword spotting, radar-based human activity recognition, and neural control of a robotic prosthesis.
Selected work.
All publicationsCleanUMamba: A Compact Mamba Network for Speech Denoising using Channel Pruning
accepted to 2025 IEEE International Symposium on Circuits and Systems (ISCAS).
SlimTTS: Parameter- and Compute-Efficient Neural Text-to-Speech Synthesis for Real-Time Inference
2026 IEEE International Symposium on Circuits and Systems (ISCAS), Shanghai, China, 2026.
DeltaKWS: A 65nm 36nJ/Decision Bio-inspired Temporal-Sparsity-Aware Digital Keyword Spotting IC with 0.6V Near-Threshold SRAM
in IEEE Transactions on Circuits and Systems for Artificial Intelligence (TCASAI), 2025
AS-ASR: A Lightweight Framework for Aphasia-Specific Automatic Speech Recognition
accepted to 2025 IEEE Biomedical Circuits and Systems (BioCAS) Conference, Abu Dhabi, United Arab Emirates, 2025
Recurrent Neural Network Control of a Hybrid Dynamical Transfemoral Prosthesis with EdgeDRNN Accelerator
2020 IEEE International Conference on Robotics and Automation (ICRA), 2020, pp. 5460-5466.