Zhuoya Zhao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhuoya Zhao has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Feifei Zhao
Brain-inspired Cognitive Intelligence Lab, Institute of Automation, Chinese Academy of Sciences, Beijing, China
4 shared publications Yi Zeng
Brain-inspired Cognitive Intelligence Lab, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
3 shared publications- Yi Zeng
Brain-inspired Cognitive AI Lab, Institute of Automation, Chinese Academy of Sciences, Beijing, China
2 shared publications Yinqian Sun
Brain-inspired Cognitive Intelligence Lab, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
2 shared publicationsBin Han
Chinese Academy of Sciences (CAS) - Research Center for Brain-inspired Intelligence
1 shared publicationFeifei Zhao
Brain-inspired Cognitive Intelligence Lab, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
1 shared publication
Latest Publications
- 31 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 5 (11) , 5760-5770
Brain-Inspired Evolutionary Architectures for Spiking Neural Networks
The intricate and distinctive evolutionary topology of the human brain enables it to execute multiple cognitive tasks simultaneously, and this automated evolutionary process of biological networks motivates our investigation into efficient architecture optimization for Spiking Neural Networks (SNNs). Diverging from traditional manual-designed and hierarchical Network Architecture Search (NAS), we advance the evolution of SNN architecture by integrating local, brain region-inspired modular structures with global cross-module connectivity. Locally, the brain region-inspired module consists of multiple neural motifs with excitatory and inhibitory connections; Globally, free connections among modules, including long-term cross-module feedforward and feedback connections are evolved. We introduce an efficient multi-objective evolutionary algorithm that leverages a few-shot predictor, endowing SNNs with high performance and low energy consumption. Extensive experiments across both static (CIFAR10, CIFAR100) and neuromorphic (CIFAR10-DVS, DVS128-Gesture) datasets reveal that the proposed model significantly exhibits robustness while maintaining consistent and exceptional performance. This study pioneers in searching for optimal neural architectures for SNNs by integrating the human brain’s advanced connectivity and modular organization into SNN optimization, thereby contributing valuable perspectives to the development of brain-inspired artificial intelligence.
- 31 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 5 (11) , 5760-5770
Brain-Inspired Evolutionary Architectures for Spiking Neural Networks
The intricate and distinctive evolutionary topology of the human brain enables it to execute multiple cognitive tasks simultaneously, and this automated evolutionary process of biological networks motivates our investigation into efficient architecture optimization for Spiking Neural Networks (SNNs). Diverging from traditional manual-designed and hierarchical Network Architecture Search (NAS), we advance the evolution of SNN architecture by integrating local, brain region-inspired modular structures with global cross-module connectivity. Locally, the brain region-inspired module consists of multiple neural motifs with excitatory and inhibitory connections; Globally, free connections among modules, including long-term cross-module feedforward and feedback connections are evolved. We introduce an efficient multi-objective evolutionary algorithm that leverages a few-shot predictor, endowing SNNs with high performance and low energy consumption. Extensive experiments across both static (CIFAR10, CIFAR100) and neuromorphic (CIFAR10-DVS, DVS128-Gesture) datasets reveal that the proposed model significantly exhibits robustness while maintaining consistent and exceptional performance. This study pioneers in searching for optimal neural architectures for SNNs by integrating the human brain’s advanced connectivity and modular organization into SNN optimization, thereby contributing valuable perspectives to the development of brain-inspired artificial intelligence.
- 3 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 5 (12) , 6280-6289
A Brain-Inspired Theory of Collective Mind Model for Efficient Social Cooperation
Social intelligence manifests the capability, often referred to as the Theory of Mind (ToM), to discern others’ behavioral intentions, beliefs, and other mental states. ToM is crucial in multi-agent and human-machine interaction contexts, where each participant need grasp the mental states of others to respond, interact, and collaborate more effectively. Recent studies show that while the ToM model can infer beliefs, intentions, and predict future observations and actions, its application in complex tasks is significantly constrained. The challenges arise when the number of agents increases, the environment becomes more complex, and interacting with the environment and predicting the mental state of each other becomes difficult and time consuming. To overcome such limits, we take inspiration from the Theory of Collective Mind (ToCM) mechanism, predicting observations of all other agents into a unified but plural representation and discerning how our own actions affect this mental state representation. Based on this foundation, we construct an imaginative space to simulate the multi-agent interaction process, thus improving the efficiency of cooperation among multiple agents in complex decision-making environments. In various cooperative tasks with different numbers of agents, the experimental results highlight the superior cooperative efficiency and performance of our approach compared to the Multi-Agent Reinforcement Learning (MARL) baselines. We achieve consistent boost on SNN- and ANN-based decision networks, and demonstrate that ToCM’s inferences about others’ mental states can be transferred to new tasks for quickly and flexible adaptation.
- 3 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 5 (12) , 6280-6289
A Brain-Inspired Theory of Collective Mind Model for Efficient Social Cooperation
Social intelligence manifests the capability, often referred to as the Theory of Mind (ToM), to discern others’ behavioral intentions, beliefs, and other mental states. ToM is crucial in multi-agent and human-machine interaction contexts, where each participant need grasp the mental states of others to respond, interact, and collaborate more effectively. Recent studies show that while the ToM model can infer beliefs, intentions, and predict future observations and actions, its application in complex tasks is significantly constrained. The challenges arise when the number of agents increases, the environment becomes more complex, and interacting with the environment and predicting the mental state of each other becomes difficult and time consuming. To overcome such limits, we take inspiration from the Theory of Collective Mind (ToCM) mechanism, predicting observations of all other agents into a unified but plural representation and discerning how our own actions affect this mental state representation. Based on this foundation, we construct an imaginative space to simulate the multi-agent interaction process, thus improving the efficiency of cooperation among multiple agents in complex decision-making environments. In various cooperative tasks with different numbers of agents, the experimental results highlight the superior cooperative efficiency and performance of our approach compared to the Multi-Agent Reinforcement Learning (MARL) baselines. We achieve consistent boost on SNN- and ANN-based decision networks, and demonstrate that ToCM’s inferences about others’ mental states can be transferred to new tasks for quickly and flexible adaptation.
- Open Access
A brain-inspired theory of mind spiking neural network improves multi-agent cooperation and competition
- Open Access
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Brain-Inspired Theory of Mind Spiking Neural Network Elevates Multi-Agent Cooperation and Competition
Theory of mind (ToM), a kind of high social cognitive ability, enables individuals to infer others’ mental states and thus explain and predict others’ behaviors. During dynamic social interaction, inferring and predicting others’ behaviors through ToM is crucial for obtaining more benefits in cooperative and competitive tasks. However, most existing multi-agent decision-making methods select behaviors based on agents’ observations, lacking in-depth inspiration from the ToM mechanism in the brain and thus limiting the performance of multi-agent decision-making. In this paper, we proposed a Multi-Agent Theory of Mind Spiking Neural Network (MAToM-SNN) model, which consists of a ToM model and a decision-making module. We designed two brain-inspired ToM modules (Self-ToM and Other-ToM) to predict others’ behavior based on self-experience and observations of others, respectively. Then, each agent will adjust its behavior according to the predicted actions of others. The proposed model has been verified in multi-agent decision-making environments with cooperative and competitive tasks. The experimental results show that introducing ToM and SNNs could improve cooperation efficiency and help achieve higher rewards compared to multi-agent reinforcement learning models. Especially for competitive tasks, the agents with ToM could improve the rewards of their team by predicting the actions of other agents.
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Open AccessBrain-Inspired Theory of Mind Spiking Neural Network Elevates Multi-Agent Cooperation and Competition
Theory of mind (ToM), a kind of high social cognitive ability, enables individuals to infer others’ mental states and thus explain and predict others’ behaviors. During dynamic social interaction, inferring and predicting others’ behaviors through ToM is crucial for obtaining more benefits in cooperative and competitive tasks. However, most existing multi-agent decision-making methods select behaviors based on agents’ observations, lacking in-depth inspiration from the ToM mechanism in the brain and thus limiting the performance of multi-agent decision-making. In this paper, we proposed a Multi-Agent Theory of Mind Spiking Neural Network (MAToM-SNN) model, which consists of a ToM model and a decision-making module. We designed two brain-inspired ToM modules (Self-ToM and Other-ToM) to predict others’ behavior based on self-experience and observations of others, respectively. Then, each agent will adjust its behavior according to the predicted actions of others. The proposed model has been verified in multi-agent decision-making environments with cooperative and competitive tasks. The experimental results show that introducing ToM and SNNs could improve cooperation efficiency and help achieve higher rewards compared to multi-agent reinforcement learning models. Especially for competitive tasks, the agents with ToM could improve the rewards of their team by predicting the actions of other agents.
- Open Access
- 1 January 2021
- Published by Elsevier in SSRN Electronic Journal
Bio-Inspired Self-Organizing Collision Avoidance for Drones Swarm Based on Reward-Modulated Spiking Neural Network
Biological systems can emerge intelligent swarm behavior through relatively simple individual, local interaction, and decentralized self-organizing decisionmaking. A major research challenge of self-organized swarm intelligence is the coupling influences between individual behaviors. Numerous swarm decisionmaking methods optimize the behavior of multiple individuals at the same time from a global perspective. Therefore, these methods are difficult to be used for online decision-making tasks of the real multi-robots system. This paper proposes a self-organized collision avoidance model for real drones incorporating a bio-inspired reward-modulated spiking neural network. Each individual independently uses the brain-inspired spiking neural network to update its own strategy according to the behavior of other agents. Subsequently, the local interaction and autonomous learning of a single individual lead to the emergence of swarm intelligence. We validated our model on swarm collision avoidance tasks in a bounded space, carrying out simulation and real-world experiments with a self-organized swarm of drones. Experimental results demonstrated that the swarm of UAVs without central control can exhibit collision avoidance even around perturbing obstacles.
- 1 January 2021
- Published by Elsevier in SSRN Electronic Journal
Open AccessBio-Inspired Self-Organizing Collision Avoidance for Drones Swarm Based on Reward-Modulated Spiking Neural Network
Biological systems can emerge intelligent swarm behavior through relatively simple individual, local interaction, and decentralized self-organizing decisionmaking. A major research challenge of self-organized swarm intelligence is the coupling influences between individual behaviors. Numerous swarm decisionmaking methods optimize the behavior of multiple individuals at the same time from a global perspective. Therefore, these methods are difficult to be used for online decision-making tasks of the real multi-robots system. This paper proposes a self-organized collision avoidance model for real drones incorporating a bio-inspired reward-modulated spiking neural network. Each individual independently uses the brain-inspired spiking neural network to update its own strategy according to the behavior of other agents. Subsequently, the local interaction and autonomous learning of a single individual lead to the emergence of swarm intelligence. We validated our model on swarm collision avoidance tasks in a bounded space, carrying out simulation and real-world experiments with a self-organized swarm of drones. Experimental results demonstrated that the swarm of UAVs without central control can exhibit collision avoidance even around perturbing obstacles.