Wooyoung Jo

About

Wooyoung Jo is a pioneering researcher in energy-efficient hardware acceleration for deep reinforcement learning (DRL), with a focus on enabling real-time, on-device training. His major contributions center on the OmniDRL processor, a groundbreaking design that achieves an impressive 29.3 TFLOPS/W efficiency—a record for DRL systems. Jo’s key innovations include dual-mode weight compression and a sparse weight transposer, which dramatically reduce memory and computation overhead by compressing weights and feature maps during every training iteration. His work addresses the critical challenge of deploying DRL on edge devices, where power and resources are limited, yet autonomous adaptation is essential. With his most-cited paper garnering 17 citations, Jo’s research has laid the foundation for practical, low-power DRL accelerators. Notably, his FPGA-based accelerator with selective mixed-precision retraining further demonstrates his commitment to online fast adaptation. By bridging the gap between algorithmic complexity and hardware efficiency, Wooyoung Jo is shaping the future of intelligent edge computing.

Research Focus

Key Achievements

Top Papers

1. 1
   OmniDRL: A 29.3 TFLOPS/W Deep Reinforcement Learning Processor with Dualmode Weight Compression and On-chip Sparse Weight Transposer

   17 citations · 2021
2. 2
   OmniDRL: An Energy-Efficient Deep Reinforcement Learning Processor With Dual-Mode Weight Compression and Sparse Weight Transposer

   4 citations · 2022
3. 3
   An Energy-Efficient Deep Reinforcement Learning FPGA Accelerator for Online Fast Adaptation with Selective Mixed-precision Re-training

   3 citations · 2021

Key Collaborators

Contact & Links