Yajun Zhu
Papers
1
Total Citations
6
H-Index
1
About
Yajun Zhu is a leading researcher in energy-efficient edge AI hardware, specializing in low-power accelerator architectures for real-time, resource-constrained environments. His most impactful work centers on dynamic reconfiguration and streaming-based convolution engines, which dramatically reduce power consumption while maintaining high inference throughput. His landmark 2023 paper, "DycSe: A Low-Power, Dynamic Reconfiguration Column Streaming-Based Convolution Engine for Resource-Aware Edge AI Accelerators," has already garnered 6 citations, reflecting its timely relevance. In this work, Zhu introduces a column-streaming dataflow that minimizes memory access and enables on-the-fly reconfiguration, allowing a single accelerator to adapt to varying neural network topologies without the overhead of traditional GPUs or general-purpose processors. This innovation is critical for deploying AI in power-constrained devices like drones, remote sensing satellites, and robotic sensors. By demonstrating that specialized hardware can outperform general-purpose solutions in both energy efficiency and latency, Zhu has established a new paradigm for resource-aware edge computing. His contributions are shaping the next generation of intelligent, autonomous systems that must operate within strict power budgets.
Research Focus
Key Achievements
Top Papers
- 1