Papers
1
Total Citations
3
H-Index
1
About
Xiaoti Wu is a rising researcher in the field of energy-efficient deep neural network (DNN) accelerators, with a focus on enabling multitasking for edge devices like intelligent robotics and autonomous vehicles. Their key research areas include memory-computing architectures, dataflow optimization, and adaptive hardware design for heterogeneous DNN workloads. Wu’s major contribution, detailed in their 2022 paper “Memory-Computing Decoupling: A DNN Multitasking Accelerator With Adaptive Data Arrangement,” introduces a novel architecture that decouples memory access from computation to handle the distinct dataflow preferences of each DNN layer in multitasking scenarios. This work addresses a critical bottleneck in real-world applications where multiple DNNs must run concurrently on resource-constrained edge platforms. While still early in their career, with 3 citations to date, Wu’s research demonstrates a forward-thinking approach to hardware-software co-design, offering a scalable solution for adaptive data arrangement that reduces memory contention and improves throughput. Their work is notable for bridging the gap between theoretical dataflow models and practical accelerator implementations, making it a valuable reference for students and researchers exploring next-generation edge AI systems.
Research Focus
Key Achievements
Top Papers
- 1