In this paper, the reinforcement learning method for cooperative multi-agent systems(MAS) with incremental number of agents is studied. The existing multi-agent reinforcement learning approaches deal with the MAS with...In this paper, the reinforcement learning method for cooperative multi-agent systems(MAS) with incremental number of agents is studied. The existing multi-agent reinforcement learning approaches deal with the MAS with a specific number of agents, and can learn well-performed policies. However, if there is an increasing number of agents, the previously learned in may not perform well in the current scenario. The new agents need to learn from scratch to find optimal policies with others,which may slow down the learning speed of the whole team. To solve that problem, in this paper, we propose a new algorithm to take full advantage of the historical knowledge which was learned before, and transfer it from the previous agents to the new agents. Since the previous agents have been trained well in the source environment, they are treated as teacher agents in the target environment. Correspondingly, the new agents are called student agents. To enable the student agents to learn from the teacher agents, we first modify the input nodes of the networks for teacher agents to adapt to the current environment. Then, the teacher agents take the observations of the student agents as input, and output the advised actions and values as supervising information. Finally, the student agents combine the reward from the environment and the supervising information from the teacher agents, and learn the optimal policies with modified loss functions. By taking full advantage of the knowledge of teacher agents, the search space for the student agents will be reduced significantly, which can accelerate the learning speed of the holistic system. The proposed algorithm is verified in some multi-agent simulation environments, and its efficiency has been demonstrated by the experiment results.展开更多
Knowledge transfer within university-led innovative research teams helps to maximally gather knowledge sources and promote knowledge dissemination,exchange and digestion among different disciplines. T he effect of tra...Knowledge transfer within university-led innovative research teams helps to maximally gather knowledge sources and promote knowledge dissemination,exchange and digestion among different disciplines. T he effect of transfer directly affects the team's capacity of knowledge innovation and its outcomes. In this paper,a WSB-based research framework for the influencing factors of knowledge transfer within university-led innovative research teams is established by means of grounded theory with help of in-depth interviews,in which five fundamental categories that affect knowledge transfer within teams,namely,knowledge source,knowledge receiver,knowledge transfer context,knowledge characteristics and knowledge transfer medium,are proposed to elaborate on the relationship between the fundamental categories and the effect of knowledge transfer within teams.Finally,a theoretical saturation test is conducted to verify the rationality and scientific tenability of this theoretical framework.展开更多
Knowledge transfer is widely emphasized as a strategic issue for firm competition. A model for intra-firm horizontal knowledge transfer is proposed to model horizontal knowledge transfer to solve some demerits in curr...Knowledge transfer is widely emphasized as a strategic issue for firm competition. A model for intra-firm horizontal knowledge transfer is proposed to model horizontal knowledge transfer to solve some demerits in current knowledge transfer researches. The concept model of intra-firm horizontal knowledge transfer was described and a framework was provided to define the main components of the transfer process. Horizontal knowledge transfer is that knowledge is transferred from the source to the same hierarchical level recipients as the target. Horizontal knowledge transfer constitutes a strategic area of knowledge management research. However, little is known about the circumstances under which one particular mechanism is the most appropriate. To address these issues, some significant conclusions are drawn concerning knowledge transfer mechanisms in a real-world setting.展开更多
A study on knowledge transfer in a mutli-agent organization is performed by applying the basic principle in physics such as the kinetic theory.Based on the theoretical analysis of the knowledge accumulation process an...A study on knowledge transfer in a mutli-agent organization is performed by applying the basic principle in physics such as the kinetic theory.Based on the theoretical analysis of the knowledge accumulation process and knowledge transfer attributes,a special type of knowledge field(KF)is introduced and the knowledge diffusion equation(KDE)is developed.The evolution of knowledge potential is modeled by lattice kinetic equation and verified by numerical experiments.The new equation-based modeling developed in this paper is meaningful to simulate and predict the knowledge transfer process in firms.The development of the lattice kinetic model(LKM)for knowledge transfer can contribute to the knowledge management theory,and the managers can also simulate the knowledge accumulation process by using the LKM.展开更多
In this paper,we study scene image recognition with knowledge transfer for drone navigation.We divide navigation scenes into three macro-classes,namely outdoor special scenes(OSSs),the space from indoors to outdoors o...In this paper,we study scene image recognition with knowledge transfer for drone navigation.We divide navigation scenes into three macro-classes,namely outdoor special scenes(OSSs),the space from indoors to outdoors or from outdoors to indoors transitional scenes(TSs),and others.However,there are difficulties in how to recognize the TSs,to this end,we employ deep convolutional neural network(CNN)based on knowledge transfer,techniques for image augmentation,and fine tuning to solve the issue.Moreover,there is still a novelty detection prob-lem in the classifier,and we use global navigation satellite sys-tems(GNSS)to solve it in the prediction stage.Experiment results show our method,with a pre-trained model and fine tun-ing,can achieve 91.3196%top-1 accuracy on Scenes21 dataset,paving the way for drones to learn to understand the scenes around them autonomously.展开更多
基金supported by the National Key R&D Program of China (2018AAA0101400)the National Natural Science Foundation of China (62173251+3 种基金61921004U1713209)the Natural Science Foundation of Jiangsu Province of China (BK20202006)the Guangdong Provincial Key Laboratory of Intelligent Decision and Cooperative Control。
文摘In this paper, the reinforcement learning method for cooperative multi-agent systems(MAS) with incremental number of agents is studied. The existing multi-agent reinforcement learning approaches deal with the MAS with a specific number of agents, and can learn well-performed policies. However, if there is an increasing number of agents, the previously learned in may not perform well in the current scenario. The new agents need to learn from scratch to find optimal policies with others,which may slow down the learning speed of the whole team. To solve that problem, in this paper, we propose a new algorithm to take full advantage of the historical knowledge which was learned before, and transfer it from the previous agents to the new agents. Since the previous agents have been trained well in the source environment, they are treated as teacher agents in the target environment. Correspondingly, the new agents are called student agents. To enable the student agents to learn from the teacher agents, we first modify the input nodes of the networks for teacher agents to adapt to the current environment. Then, the teacher agents take the observations of the student agents as input, and output the advised actions and values as supervising information. Finally, the student agents combine the reward from the environment and the supervising information from the teacher agents, and learn the optimal policies with modified loss functions. By taking full advantage of the knowledge of teacher agents, the search space for the student agents will be reduced significantly, which can accelerate the learning speed of the holistic system. The proposed algorithm is verified in some multi-agent simulation environments, and its efficiency has been demonstrated by the experiment results.
基金Project supported by the MOE Planned Fund for Humanities and Social Sciences(Project Name:Empirical Research into the Influencing Factors of Knowledge Transfer within University-led Innovative Research TeamsGrant No.:12YJA630169)
文摘Knowledge transfer within university-led innovative research teams helps to maximally gather knowledge sources and promote knowledge dissemination,exchange and digestion among different disciplines. T he effect of transfer directly affects the team's capacity of knowledge innovation and its outcomes. In this paper,a WSB-based research framework for the influencing factors of knowledge transfer within university-led innovative research teams is established by means of grounded theory with help of in-depth interviews,in which five fundamental categories that affect knowledge transfer within teams,namely,knowledge source,knowledge receiver,knowledge transfer context,knowledge characteristics and knowledge transfer medium,are proposed to elaborate on the relationship between the fundamental categories and the effect of knowledge transfer within teams.Finally,a theoretical saturation test is conducted to verify the rationality and scientific tenability of this theoretical framework.
文摘Knowledge transfer is widely emphasized as a strategic issue for firm competition. A model for intra-firm horizontal knowledge transfer is proposed to model horizontal knowledge transfer to solve some demerits in current knowledge transfer researches. The concept model of intra-firm horizontal knowledge transfer was described and a framework was provided to define the main components of the transfer process. Horizontal knowledge transfer is that knowledge is transferred from the source to the same hierarchical level recipients as the target. Horizontal knowledge transfer constitutes a strategic area of knowledge management research. However, little is known about the circumstances under which one particular mechanism is the most appropriate. To address these issues, some significant conclusions are drawn concerning knowledge transfer mechanisms in a real-world setting.
基金supported by the National Natural Science Foundation of China(71472055 71871007)+2 种基金National Social Science Foundation of China(16AZD0006)Heilongjiang Philosophy and Social Science Research Project(19GLB087)the Fundamental Research Funds for the Central Universities(HIT.NSRIF.2019033)
文摘A study on knowledge transfer in a mutli-agent organization is performed by applying the basic principle in physics such as the kinetic theory.Based on the theoretical analysis of the knowledge accumulation process and knowledge transfer attributes,a special type of knowledge field(KF)is introduced and the knowledge diffusion equation(KDE)is developed.The evolution of knowledge potential is modeled by lattice kinetic equation and verified by numerical experiments.The new equation-based modeling developed in this paper is meaningful to simulate and predict the knowledge transfer process in firms.The development of the lattice kinetic model(LKM)for knowledge transfer can contribute to the knowledge management theory,and the managers can also simulate the knowledge accumulation process by using the LKM.
基金supported by the National Natural Science Foundation of China(62103104)the Natural Science Foundation of Jiangsu Province(BK20210215)the China Postdoctoral Science Foundation(2021M690615).
文摘In this paper,we study scene image recognition with knowledge transfer for drone navigation.We divide navigation scenes into three macro-classes,namely outdoor special scenes(OSSs),the space from indoors to outdoors or from outdoors to indoors transitional scenes(TSs),and others.However,there are difficulties in how to recognize the TSs,to this end,we employ deep convolutional neural network(CNN)based on knowledge transfer,techniques for image augmentation,and fine tuning to solve the issue.Moreover,there is still a novelty detection prob-lem in the classifier,and we use global navigation satellite sys-tems(GNSS)to solve it in the prediction stage.Experiment results show our method,with a pre-trained model and fine tun-ing,can achieve 91.3196%top-1 accuracy on Scenes21 dataset,paving the way for drones to learn to understand the scenes around them autonomously.
