Zixuan Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zixuan Wang has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Ganggang Dong
National Key Laboratory of Radar Signal Processing, Xidian University, Xian, China
1 shared publication Gongmin Li
School of Integrated Circuits Huazhong University of Science and Technology Wuhan 430074 China
1 shared publication- Hongwei Liu
National Key Laboratory of Radar Signal Processing, Xidian University, Xian, China
1 shared publication Jian Xia
School of Integrated Circuits Huazhong University of Science and Technology Wuhan 430074 China
1 shared publicationJie Gong
School of Integrated Circuits Huazhong University of Science and Technology Wuhan 430074 China
1 shared publication- Rui Yang
Huazhong University of Science and Technology
1 shared publication
Latest Publications
- 5 August 2025
- Published by Wiley in Laser & Photonics Reviews
- Vol. 19 (24)
Spike‐Timing‐Dependent Plasticity Realized in the Silicon Photonic Synapse Based on Phase Change Materials
Optical spiking neural networks (OSNNs) based on phase‐change materials (PCMs) synergistically integrate the energy‐efficient nature of spiking neural networks, the low‐latency, high‐bandwidth advantages offered by optical neural networks, with high optical contrast, nonvolatile storage of PCM, emerging as a pivotal technology to address data redundancy and elevated energy consumption challenges in the information processing. The synaptic connections within OSNNs are modulated by spike‐timing‐dependent plasticity (STDP) learning rules. However, the STDP learning rules are not realized in PCM‐based electrically programmable photonic devices, due to the challenge of precisely controlling the PCM modulation. Here, a Ge 2 Sb 2 Te 5 ‐based photonic device with the implementation of four STDP learning rules is demonstrated, enabled by the introduction of an Al 2 O 3 thermal conductive layer. The present photonic device shows outstanding performances, including an endurance of over 45000 switching cycles and 70 intermediate crystalline states, as well as 25 intermediate amorphous states. The OSNN employing the present device as synapses achieves an accuracy of up to 94.07% in the recognition task, outperforming counterparts based on unilateral modulation devices. This investigation indicates that the integration of phase‐change photonic devices with OSNN shows great potential for building the next generation of neuromorphic computing systems with high energy efficiency.
- 5 August 2025
- Published by Wiley in Laser & Photonics Reviews
- Vol. 19 (24)
Spike‐Timing‐Dependent Plasticity Realized in the Silicon Photonic Synapse Based on Phase Change Materials
Optical spiking neural networks (OSNNs) based on phase‐change materials (PCMs) synergistically integrate the energy‐efficient nature of spiking neural networks, the low‐latency, high‐bandwidth advantages offered by optical neural networks, with high optical contrast, nonvolatile storage of PCM, emerging as a pivotal technology to address data redundancy and elevated energy consumption challenges in the information processing. The synaptic connections within OSNNs are modulated by spike‐timing‐dependent plasticity (STDP) learning rules. However, the STDP learning rules are not realized in PCM‐based electrically programmable photonic devices, due to the challenge of precisely controlling the PCM modulation. Here, a Ge 2 Sb 2 Te 5 ‐based photonic device with the implementation of four STDP learning rules is demonstrated, enabled by the introduction of an Al 2 O 3 thermal conductive layer. The present photonic device shows outstanding performances, including an endurance of over 45000 switching cycles and 70 intermediate crystalline states, as well as 25 intermediate amorphous states. The OSNN employing the present device as synapses achieves an accuracy of up to 94.07% in the recognition task, outperforming counterparts based on unilateral modulation devices. This investigation indicates that the integration of phase‐change photonic devices with OSNN shows great potential for building the next generation of neuromorphic computing systems with high energy efficiency.
- 26 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
A Cross-Modality Contrastive Learning Method for Radar Jamming Recognition
The discernment of jamming signal was important for the downstream tasks. Though great improvement was achieved by deep learning previously, these methods need large amounts of signals with label information. It was yet difficult to be met in the practical applications. To solve the problem, a cross-modality contrastive learning method is proposed in this paper. The signal-wise hierarchy and the image-wise hierarchy were presented to learn the features from IQ tensor and TF-image, respectively. The cross-domain features were then aggregated subsequently. The obtained features were delivered to the learning architecture. It was composed of the pre-training and the fine-tuning. The signals without label information were first used to pre-train a base model. The similarity loss that made the positive more similar to the signal than the negative was presented to optimize the model. The pre-trained model was then fine-tuned by a small amounts of labeled signals. The recognition task can be then achieved accordingly. So the labeled signals and the unlabeled ones were jointly exploited. Likewise, two different kinds of modal data were unified into a single framework. Multiple rounds of experiments were finally performed. The results thrown light on the superiority of proposed method over the standard and recent techniques.
- 26 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
A Cross-Modality Contrastive Learning Method for Radar Jamming Recognition
The discernment of jamming signal was important for the downstream tasks. Though great improvement was achieved by deep learning previously, these methods need large amounts of signals with label information. It was yet difficult to be met in the practical applications. To solve the problem, a cross-modality contrastive learning method is proposed in this paper. The signal-wise hierarchy and the image-wise hierarchy were presented to learn the features from IQ tensor and TF-image, respectively. The cross-domain features were then aggregated subsequently. The obtained features were delivered to the learning architecture. It was composed of the pre-training and the fine-tuning. The signals without label information were first used to pre-train a base model. The similarity loss that made the positive more similar to the signal than the negative was presented to optimize the model. The pre-trained model was then fine-tuned by a small amounts of labeled signals. The recognition task can be then achieved accordingly. So the labeled signals and the unlabeled ones were jointly exploited. Likewise, two different kinds of modal data were unified into a single framework. Multiple rounds of experiments were finally performed. The results thrown light on the superiority of proposed method over the standard and recent techniques.