Zirui Zhao

Dr. Zirui Zhao

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 December 2024 in Electronics
Verified
Journal Article
Open access
PDF file
SGShouwei Gao
Shouwei Gao
YQYu Qin
Yu Qin
RZRuixin Zhu
Ruixin Zhu
ZZZirui Zhao
Zirui Zhao
HZHao Zhou
Hao Zhou
ZZZihao Zhu
Zihao Zhu
Spiking neural networks (SNNs), a neural network model structure inspired by the human brain, have emerged as a more energy-efficient deep learning paradigm due to their unique spike-based transmission and event-driven characteristics. Combining SNNs with the Transformer model significantly enhances SNNs’ performance while maintaining good energy efficiency. The gating mechanism, which dynamically adjusts input data and controls information flow, plays an important role in artificial neural networks (ANNs). Here, we introduce this gating mechanism into SNNs and propose a novel spike Transformer model, called SGSAFormer, based on the Spikformer network architecture. We introduce the Spike Gated Linear Unit (SGLU) module to improve the Multi-layer perceptron (MLP) module in SNNs by adding a gating mechanism to enhance the model’s expressive power. We also incorporate Spike Gated Self-Attention (SGSA) to strengthen the network’s attention mechanism, improving its ability to capture temporal information and dynamic processing. Additionally, we propose a Temporal Attention (TA) module, which selects new filters for the input data along the temporal dimension and can substantially reduce energy consumption with only a slight decrease in accuracy. To validate the effectiveness of our approach, we conducted extensive experiments on several neuromorphic datasets. Our model outperforms other state-of-the-art models in terms of performance.

31 December 2025 in Electronics
Verified
Journal Article
Open access
PDF file
JLJiao Li
Jiao Li
ZZZirui Zhao
Zirui Zhao
SGShouwei Gao
Shouwei Gao
SRSijie Ran
Sijie Ran
Spiking Neural Networks (SNNs) offer promising low-power alternatives to conventional neural models but often incur considerable redundancy in parameters and computations. To address these inefficiencies, we propose SpikingDynamicMaskFormer (SDMFormer), a novel framework that integrates dynamic masking and lightweight position encoding into a spike-based Transformer backbone. Specifically, our Dynamic Mask Encoder Block adaptively suppresses ineffective spike channels by learning mask parameters, reducing parameter count to 37.93–42.69% of the original Spikformer. Simultaneously, a redesigned lightweight position embedding replaces resource-intensive relative position convolutions, further lowering complexity. Experiments on three neuromorphic vision datasets—DVS128, CIFAR10-DVS and N-Caltech101—demonstrate that SDMFormer cuts energy consumption by 42.79–50.13% relative to Spikformer while maintaining or slightly surpassing accuracy. Moreover, compared with recent leading works, SDMFormer achieves competitive accuracy with substantially fewer parameters and delivers higher inference efficiency, reaching up to 196.20 img/s on CIFAR10-DVS. These results highlight the efficacy of combining event-driven attention with structured pruning and parameter-efficient position encoding, indicating the potential of SDMFormer for resource-efficient SNN deployment in low-power applications.