Biography
Xinming Shi received the B.E. degree in electronic engineering from Wuhan University of Technology and B.A. degree in English literature from from Wuhan University, Wuhan, China, in 2016. She received the M.E. degree in School of Artificial Intelligence & Automation from Huazhong University of Science and Technology, Wuhan, China, in 2019. In 2023, she got her Ph.D. degree in Computer Science at the University of Birmingham, U.K., in collaboration with Southern University of Science and Technology, Shenzhen, China. She has spent her Ph.D from three different universities during Ph.D., including UoB in UK, Southern University of Science and Technology (SUSTech) in China, and Victoria University of Wellington (VUW) in New Zealand. She has published more than a dozen papers, including those in leading IEEE Transactions and ACM Transactions.
Research Interests
Inspired by the human brain, the development of brain-inspired intelligence is a rapidly expanding area in AI. Facing the constraints of traditional von Neumann architectures and the diminishing of Moore's Law, the computing performance of the intelligent hardware needs to be continuously improved. We delve into devices, algorithms, circuits, and architecture, focusing on the use of brain-inspired learning, memory, and evolution in designing software algorithms and hardware, targeting applications in edge computing, robotics, and large language models.
- Neuromorphic Computing Hardware & Software
- Evolutionary Learning
- Trustworthy Brain-inspired Intelligence
Open Positions
We are actively seeking individuals with strong programming and circuit design skills to tackle challenges at the intersection of neuromorphic computing and computational intelligence. If you are interested, we have multiple PhD, Joint PhD, Part-time PhD, RA, and Visiting Scholar positions available. Our research focuses on the following prospective topics:
- Brain-inspired intelligence algorithms, such as spiking neural networks
- Design of trustworthy brain-inspired computing systems
- Evolutionary learning algorithm and its brain-inspired hardware implementation
- Brain-inspired hardware-algorithm co-design framework
- Efficient brain-inspired LLM algorithm/hardware design
If you are interested, please submit your CV to my email.
Research Samples
A Brain-Inspired Hardware Architecture for Evolutionary Algorithms Based on Memristive Arrays
Zilu Wang, Xinming Shi, and Xin Yao
This work introduces a hardware architecture for evolutionary algorithms using memristive arrays, enabling sparse and approximate computing through parallel analog computation. Our implementation demonstrates a significant speed improvement of at least four orders of magnitude. Additionally, we investigate fault tolerance and parameter adaptability through grounded simulations. The results show that the system can evolve and adapt to failures or changes in the memristive arrays, highlighting the architecture's robustness and adaptability.
Zilu Wang, Xinming Shi, and Xin Yao
This work introduces a hardware architecture for evolutionary algorithms using memristive arrays, enabling sparse and approximate computing through parallel analog computation. Our implementation demonstrates a significant speed improvement of at least four orders of magnitude. Additionally, we investigate fault tolerance and parameter adaptability through grounded simulations. The results show that the system can evolve and adapt to failures or changes in the memristive arrays, highlighting the architecture's robustness and adaptability.
Evolving Memristive Reservoir
Xinming Shi, Leandro L. Minku,and Xin Yao
Memristive reservoirs have gained significant attention in various research fields. However, their deterministic hardware implementation poses challenges for hardware reservoir adaptation. In this work, we introduce an evolvable memristive reservoir circuit based on reconfigurable memristive units (RMUs) that can adaptively evolve for different tasks, directly evolving the configuration signals of memristors to avoid device variance. We also propose a scalable algorithm for evolving this reconfigurable memristive reservoir circuit, ensuring circuit validity according to circuit laws and sparse topology for scalability and feasibility during evolution.
Xinming Shi, Leandro L. Minku,and Xin Yao
Memristive reservoirs have gained significant attention in various research fields. However, their deterministic hardware implementation poses challenges for hardware reservoir adaptation. In this work, we introduce an evolvable memristive reservoir circuit based on reconfigurable memristive units (RMUs) that can adaptively evolve for different tasks, directly evolving the configuration signals of memristors to avoid device variance. We also propose a scalable algorithm for evolving this reconfigurable memristive reservoir circuit, ensuring circuit validity according to circuit laws and sparse topology for scalability and feasibility during evolution.
Adaptive Memory-Enhanced Time Delay Reservoir and its Memristive Implementation (Execllent Science & Technology Academic Papers presented by SZSTA)
Xinming Shi, Leandro L. Minku,and Xin Yao
The Time Delay Reservoir (TDR) is a hardware-friendly machine learning approach that reduces connection overhead in neural networks. However, it struggles with long-term dependency tasks. We introduce a higher-order delay unit to enhance reservoir memory, using PSO for adaptivity. We also propose a memristive implementation of TDR, using dynamic memristors and memristor-based delay elements, which is feasible, effective, and more efficient in terms of circuit area and power consumption compared to traditional hardware reservoirs.
Xinming Shi, Leandro L. Minku,and Xin Yao
The Time Delay Reservoir (TDR) is a hardware-friendly machine learning approach that reduces connection overhead in neural networks. However, it struggles with long-term dependency tasks. We introduce a higher-order delay unit to enhance reservoir memory, using PSO for adaptivity. We also propose a memristive implementation of TDR, using dynamic memristors and memristor-based delay elements, which is feasible, effective, and more efficient in terms of circuit area and power consumption compared to traditional hardware reservoirs.
A Novel Tree-based Representation for Evolving Analog Circuits and Its Application to Memristor-based Pulse Generation Circuit
Xinming Shi, Leandro L. Minku,and Xin Yao
In circuit design automation using evolutionary algorithms, circuit representation is crucial. Existing studies on circuit representations face issues such as limited design diversity and inefficient SPICE netlist conversion. This paper introduces a novel tree-based representation for analog circuits, featuring an intuitive mapping rule for SPICE netlists and an effective crossover operator. We propose a genetic programming framework that evolves both circuit topology and device values. The approach is validated with three benchmark circuits and a memristor-based pulse generation circuit, showcasing its compactness and energy efficiency compared to manually-designed counterparts.
Xinming Shi, Leandro L. Minku,and Xin Yao
In circuit design automation using evolutionary algorithms, circuit representation is crucial. Existing studies on circuit representations face issues such as limited design diversity and inefficient SPICE netlist conversion. This paper introduces a novel tree-based representation for analog circuits, featuring an intuitive mapping rule for SPICE netlists and an effective crossover operator. We propose a genetic programming framework that evolves both circuit topology and device values. The approach is validated with three benchmark circuits and a memristor-based pulse generation circuit, showcasing its compactness and energy efficiency compared to manually-designed counterparts.
Memristor-Based Circuit Design for Neuron With Homeostatic Plasticity
Xinming Shi, Zhigang Zeng, Le Yang and Yi Huang
In this paper, a memristive neuron with homeostatic plasticity is implemented. The memristor is integrated into the neuron circuit, where its memristance represents the membrane sensitivity of neurons. The memristor-based neuron comprises a trigger module, a pulse generation module, and a feedback module. This configuration allows the neuron to adaptively adjust its firing rate and maintain it within the inherent range, mirroring its biological counterpart. The proposed design can replicate the firing rate behaviors observed in electrophysiological experiments on rodents, where the homeostatic plasticity of neurons is inhibited by sleep signals and promoted by wake signals. Furthermore, all simulations are conducted in Cadence PSPICE, validating the functionality of the design.
Xinming Shi, Zhigang Zeng, Le Yang and Yi Huang
In this paper, a memristive neuron with homeostatic plasticity is implemented. The memristor is integrated into the neuron circuit, where its memristance represents the membrane sensitivity of neurons. The memristor-based neuron comprises a trigger module, a pulse generation module, and a feedback module. This configuration allows the neuron to adaptively adjust its firing rate and maintain it within the inherent range, mirroring its biological counterpart. The proposed design can replicate the firing rate behaviors observed in electrophysiological experiments on rodents, where the homeostatic plasticity of neurons is inhibited by sleep signals and promoted by wake signals. Furthermore, all simulations are conducted in Cadence PSPICE, validating the functionality of the design.
Publication List
- Xingming Shi, Leandro L. Minku and Xin Yao, "Evolving Memristive Reservoir," in IEEE Transactions on Neural Networks and Learning Systems, doi: 10.1109/TNNLS.2023.3270224.
- Xingming Shi, Leandro L. Minku and Xin Yao, "Adaptive Memory-Enhanced Time Delay Reservoir and its Memristive Implementation," in IEEE Transactions on Computers, vol. 71, no. 11, pp. 2766-2777, 1 Nov. 2022, doi: 10.1109/TC.2022.3173151. (Execllent Science & Technology Academic Papers presented by SZSTA)
- Xinming Shi, Leandro L. Minku, and Xin Yao, "A Novel Tree-based Representation for Evolving Analog Circuits and Its Application to Memristor-based Pulse Generation Circuit," Genetic Programming and Evolvable Machines, 23, pp. 453–493, 2022, https://doi.org/10.1007/s10710-022-09436-w
- Xinming Shi, Zhigang Zeng, Le Yang and Yi Huang, "Memristor-Based Circuit Design for Neuron With Homeostatic Plasticity," in IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 2, no. 5, pp. 359-370, Oct. 2018, doi: 10.1109/TETCI.2018.2829914.
- Zilu Wang, Xingming Shi, and Xin Yao, "A Brain-Inspired Hardware Architecture for Evolutionary Algorithms based on Memristive Arrays," in ACM Transactions on Design Automation of Electronic Systems, doi: 10.1145/3598421.
- Xingming Shi, Leandro L. Minku and Xin Yao, "Tree-based Genetic Programming for Evolutionary Analog Circuit with Approximate Shapley Value," in AI-2024: The Forty-fourth SGAI International Conference , Cambridge, UK, 2024, accepted.
- Xingming Shi, Leandro L. Minku and Xin Yao, "Novel Memristive Reservoir Computing with Evolvable Topology for Time Series Prediction," in 31st International Conference on Neural Information Processing (ICONIP), Auckland, NZ, 2024, accepted.
- Xingming Shi, Zilu Wang, Leandro L. Minku and Xin Yao, "Explaining Memristive Reservoir Computing Through Evolving Feature Attribution," in Proceedings of the Genetic and Evolutionary Computation Conference Companion, 2023, doi: 10.1145/3583133.3590619.
- Xinming Shi, Jiashi Gao, Leandro L. Minku, and Xin Yao, “Evolving Parsimonious Circuits Through Shapley Value-based Genetic Programming,” in Proceedings of the Genetic and Evolutionary Computation Conference Companion, 2022, pp. 602–605.
- Xinming Shi, Jiashi Gao, Leandro L. Minku, James Jian Qiao Yu and Xin Yao, "Second-order Time Delay Reservoir Computing for Nonlinear Time Series Problems," 2021 IEEE Symposium Series on Computational Intelligence (SSCI), Orlando, FL, USA, 2021, pp. 1-8, doi: 10.1109/SSCI50451.2021.9659913.
- Xinming Shi and Zilu Wang, "Memristor Modeling with Homeostatic Threshold Variation for Simulation and Application," 2023 International Conference on Neuromorphic Computing (ICNC), Wuhan, China, 2024, pp. 281-290, doi: 10.1109/ICNC59488.2023.10462883.
- Le Yang, Zhigang Zeng, and Xinming Shi, “A memristor-based neural network circuit with synchronous weight adjustment,” Neurocomputing, vol. 363, pp. 114–124, 201.
- Xinming Shi and Zhigang Zeng, "Memristor-Based Neuron Circuit with Adaptive Firing Rate," 2018 Eighth International Conference on Information Science and Technology (ICIST), Cordoba, Granada, and Seville, Spain, 2018, pp. 176-181, doi: 10.1109/ICIST.2018.8426182.
- Jiashi Gao, Xinming Shi and James Jian Qiao Yu, "Attn-CommNet: Coordinated Traffic Lights Control On Large-Scale Network Level," 2021 IEEE 33rd International Conference on Tools with Artificial Intelligence (ICTAI), Washington, DC, USA, 2021, pp. 289-293, doi: 10.1109/ICTAI52525.2021.00048.
- Jiashi Gao, Xinming Shi and James Jian Qiao Yu, "Social-dualcvae: Multimodal Trajectory Forecasting Based on Social Interactions Pattern Aware and Dual Conditional Variational Auto-encoder," arXiv preprint, arXiv:2202.03954.
Patents
- Xinming Shi and Xin Yao, "基于树结构的模拟电路自动设计方法、装置、设备及介质," Pub Number: CN113420519 B, 7th Apr. 2023. (China patent)
- Xinming Shi and Xin Yao, "Automatic design method and device for analog circuit based on tree structure, equipment and medium," Pub Number: CN202110713376.7, Jun. 2021. (US patent)
- Xinming Shi and Xin Yao, "Automatisches Entwurfsvorrichtung fur eine analoge Schaltung basierend auf enier Baumstruktur," Prioritat: 25.06.2021, Sep. 2022. (German patent)
- Xinming Shi and Zhigang Zeng, “A memristor-based neuron circuit with homeostatic plasticity”, Pub Number CN107742153A, Feb. 2018. (China patent)
- Huazhong Xu, Miaoke Chen, Xinming Shi, Hang Yang, Xiao Peng and Jian Luo, “Concentrated treatment of living oil fumes emissions”, Pub Number 201530152318.7, May. 2015.
- Huazhong Xu, Yixin Wang, Xinming Shi, Xipeng Yu, Xiao Peng, Miaoke Chen, “An emission device of living oil fumes”, Pub Number 201520324700.6, May. 2015.
Research Grant
- 2023 IEEE CIS Graduate Student Research Grants, PI (USD 4,000), IEEE Computational Intelligence Society (CIS)
Honors & Awards
- 2024 Recipient of Leverhulme Early Career Fellowships, Leverhulme Trust
- 2024 President Excellent Postdoctoral, SUSTECH
- 2024 Top Ten Graduates of Southern University of Science and Technology, SUSTECH
- 2022 Excellent Science & Technology Academic Papers, Shenzhen Association for Science and Technology (SZSTA)
- 2022 Excellent Student Teaching Assistant Award, SUSTECH
- 2018 Outstanding Graduate Student Leader, HUST
- 2015 Meritorious Winner, Mathematical Contest in Modeling (MCM), Consortium for Mathematics and Its Applications (COMAP)
- 2015 First Prize, National University Student Social Practice and Science Contest on Energy Saving and Emission Reduction, WHUT
- 2014 Second Prize, The 1st Delta Advanced Automation Contest, Chinese Association of Automation
- 2013 Social Work Award, WHUT
- 2013 Outstanding Volunteer Award, The 9th Wuhan Sports Games
- 2012-2014 Merit Student Award (2 times), WHUT
- 2012 Prominent Student Award (top 5%), WHUT
- 2012 Scholarship for Outstanding Learning Achievement (top 5%), WHUT
Professional Services
Memberships
- 2022–Now: SIGEVO Member
- 2023–Now: IEEE Computational Intelligence Society (CIS) Member
- 2023–Now: Chinese Institute of Electronics Member
Committee Services
- 2021–2023: Committee Member, Conference Activities and Communications Subcommittee in IEEE Computational Intelligence Society (CIS).
Reviews
- Reviewer of TNNLS
- Reviewer of TCDS
- Reviewer of TETCI
- Reviewer of IJCNN
- PC member of LA-CCI 2023