Shengbing Zhang
Papers
1
Total Citations
3
H-Index
1
About
Shengbing Zhang is a leading researcher in energy-efficient deep neural network (DNN) accelerators, with a focus on memory-computing architectures for edge and autonomous systems. His most cited work, "Memory-Computing Decoupling: A DNN Multitasking Accelerator With Adaptive Data Arrangement" (2022), tackles the critical challenge of running multiple DNNs simultaneously on resource-constrained devices. Zhang’s key contribution lies in decoupling memory access from computation, enabling adaptive dataflows that match the heterogeneous layer preferences of different subtasks—a breakthrough for intelligent robotics and autonomous vehicles. This design significantly reduces data movement and energy consumption while maintaining high throughput. With 3 citations, this paper is gaining traction as a foundational reference for multitasking accelerator research. Zhang’s work bridges the gap between theoretical dataflow optimization and practical hardware implementation, offering a scalable solution for real-world AI workloads. His achievements highlight a deep understanding of the memory wall problem in edge computing, positioning him as an emerging innovator in the field of domain-specific architectures for AI.
Research Focus
Key Achievements
Top Papers
- 1