Kunpeng Wan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Kunpeng Wan has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Fan Xu
College of Computer and Information Technology, China Three Gorges University, Yichang 443002, China
1 shared publication - Helin Yang
Department of Information and Communication Engineering, School of Informatics, Xiamen University, Xiamen, China
1 shared publication - Liang Xiao
Xiamen University
1 shared publication - Yanyong Zhang
University of Science and Technology of China, Hefei, Anhui, China
1 shared publication - Yaqi Xu
School of Computer Science and Engineer, Central South University, Changsha, China
1 shared publication - Yi Wu
Fujian Normal University
1 shared publication
Latest Publications
- 14 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Communications
- Vol. 71 (2) , 864-876
Reinforcement Learning Based Energy-Efficient Collaborative Inference for Mobile Edge Computing
Collaborative inference in mobile edge computing (MEC) enables mobile devices to offload the computation tasks for the computation-intensive perception services, and the inference policy determines the inference latency and energy consumption. The optimal inference policy depends on the inference performance model of deep learning, the data generation model and the network model that are rarely known by mobile devices in time. In this paper, we propose a multi-agent reinforcement learning (RL) based energy-efficient MEC collaborative inference scheme, which enables each mobile device to choose both the partition point of deep learning and the collaborative edge of each mobile device based on the image quantity, the channel conditions and the previous inference performance. A learning experience exchange mechanism exploits the Q-values of the neighboring mobile devices to accelerate the inference policy optimization with less energy consumption. We also provide a deep multi-agent RL based inference scheme to accelerate learning for large-scale MEC networks, in which an actor network yields the collaborative inference policy probability distribution and a critic network guides the weight update of the actor network to enhance sample efficiency. We provide the inference performance bound and analyze the computational complexity. Both simulation and experimental results show that our proposed schemes reduce the inference latency and save the MEC energy consumption.
- 14 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Communications
- Vol. 71 (2) , 864-876
Reinforcement Learning Based Energy-Efficient Collaborative Inference for Mobile Edge Computing
Collaborative inference in mobile edge computing (MEC) enables mobile devices to offload the computation tasks for the computation-intensive perception services, and the inference policy determines the inference latency and energy consumption. The optimal inference policy depends on the inference performance model of deep learning, the data generation model and the network model that are rarely known by mobile devices in time. In this paper, we propose a multi-agent reinforcement learning (RL) based energy-efficient MEC collaborative inference scheme, which enables each mobile device to choose both the partition point of deep learning and the collaborative edge of each mobile device based on the image quantity, the channel conditions and the previous inference performance. A learning experience exchange mechanism exploits the Q-values of the neighboring mobile devices to accelerate the inference policy optimization with less energy consumption. We also provide a deep multi-agent RL based inference scheme to accelerate learning for large-scale MEC networks, in which an actor network yields the collaborative inference policy probability distribution and a critic network guides the weight update of the actor network to enhance sample efficiency. We provide the inference performance bound and analyze the computational complexity. Both simulation and experimental results show that our proposed schemes reduce the inference latency and save the MEC energy consumption.
- Open Access
- 6 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 8436-8448
A Node Selecting Approach for Traffic Network Based on Artificial Slime Mold
The node selecting problem of traffic network is a significant issue and is difficult to be solved. In this paper, an artificial slime mold method is proposed to help us solve the problem. First, the chief components of an artificial slime mold are introduced to simulate the foraging behavior of a true slime mold, including external food sources, plasmodium, myxamoeba, nucleus, and nutrients. Then the learning mechanism of nutrient concentration for the artificial slime mold is illustrated, though there is no brain or neuron in its body. After that, the node selecting approach is described according to the propagation capabilities of nodes. Second, the algorithm flow is designed to show how to solve this kind of complex selecting problem. The algorithm flow to select important traffic nodes by artificial slime mold is composed of 4 main steps, including initialization, food searching, feeding, and selecting for output. Third, a comprehensive example is designed and derived from references to certificate that the proposed artificial slime mold can help us select important traffic nodes by their generated traffic topologies. The contributions of this paper are important both for traffic node selecting and artificial learning mechanism in theoretical and practical aspects.
- 6 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 8436-8448
Open AccessA Node Selecting Approach for Traffic Network Based on Artificial Slime Mold
The node selecting problem of traffic network is a significant issue and is difficult to be solved. In this paper, an artificial slime mold method is proposed to help us solve the problem. First, the chief components of an artificial slime mold are introduced to simulate the foraging behavior of a true slime mold, including external food sources, plasmodium, myxamoeba, nucleus, and nutrients. Then the learning mechanism of nutrient concentration for the artificial slime mold is illustrated, though there is no brain or neuron in its body. After that, the node selecting approach is described according to the propagation capabilities of nodes. Second, the algorithm flow is designed to show how to solve this kind of complex selecting problem. The algorithm flow to select important traffic nodes by artificial slime mold is composed of 4 main steps, including initialization, food searching, feeding, and selecting for output. Third, a comprehensive example is designed and derived from references to certificate that the proposed artificial slime mold can help us select important traffic nodes by their generated traffic topologies. The contributions of this paper are important both for traffic node selecting and artificial learning mechanism in theoretical and practical aspects.