Dan Shen
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
Research Subjects
Dan Shen has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Xiangwei Zhu
Sun Yat-sen University
4 shared publications - Bo Xu
School of Aeronautics and Astronautics, Sun Yat-sen University, Shenzhen, China
2 shared publications - Mingjun Ouyang
School of Electronics and Information, Guangdong Polytechnic Normal University, Guangzhou, China
2 shared publications - Yang Liu
School of Electronics and Communication Engineering, Sun Yat-sen University, Shenzhen, China
2 shared publications - Bo Gao
College of Animal Science and Technology, Yangzhou University, Yangzhou 225009, China
1 shared publication - Cai Chen
College of Animal Science & Technology, Yangzhou University, Yangzhou 225009, China
1 shared publication
Latest Publications
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (11) , 19646-19656
CASIT: Collective Intelligent Agent System for Internet of Things
In the last few years, the bottleneck of bandwidth in Internet of Thing (IoT) has driven expectations to figure out new ways to preprocess the information needed to be transmitted. The ways which were used before are not smart enough and they cannot align to the users’ need. Large language model (LLM)-based intelligent agent is a very hot concept in AI community, which aims to save various problems via adapting LLM to different industries. In this article, we present a collective intelligent agent system for the IoT (CASIT) that is a pioneering LLM-agent-based IoT system. We put forward a IoT framework that can be used to lots of scenarios. CASIT refers to a system based on multiple intelligent LLM agents, which realizes complex tasks through cooperation and makes full use of collective intelligence. In order to solve the problems, we designed the Memory Mechanism and Summary Mechanism that enable LLMs to efficiently process the data by comparing historical data with Local Knowledge and Chat History in the prompt. After experimental verification, we have found that our framework could accurately conclude the abnormal information, and it outperforms the single LLM system when we input 200 sets of temperature and humidity data from five different places. The system provides a new solution and method for information processing in all IoT systems. Our framework may also provide refreshing ideas for edge computing and semantic communication.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (11) , 19646-19656
CASIT: Collective Intelligent Agent System for Internet of Things
In the last few years, the bottleneck of bandwidth in Internet of Thing (IoT) has driven expectations to figure out new ways to preprocess the information needed to be transmitted. The ways which were used before are not smart enough and they cannot align to the users’ need. Large language model (LLM)-based intelligent agent is a very hot concept in AI community, which aims to save various problems via adapting LLM to different industries. In this article, we present a collective intelligent agent system for the IoT (CASIT) that is a pioneering LLM-agent-based IoT system. We put forward a IoT framework that can be used to lots of scenarios. CASIT refers to a system based on multiple intelligent LLM agents, which realizes complex tasks through cooperation and makes full use of collective intelligence. In order to solve the problems, we designed the Memory Mechanism and Summary Mechanism that enable LLMs to efficiently process the data by comparing historical data with Local Knowledge and Chat History in the prompt. After experimental verification, we have found that our framework could accurately conclude the abnormal information, and it outperforms the single LLM system when we input 200 sets of temperature and humidity data from five different places. The system provides a new solution and method for information processing in all IoT systems. Our framework may also provide refreshing ideas for edge computing and semantic communication.
- Open Access
- 13 December 2023
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-1/W, 1141-1146
DYNAMICAL VARIATIONAL AUTOENCODERS AND KALMANNET: NEW APPROACHES TO ROBUST HIGH-PRECISION NAVIGATION
Kalman filters, recognized as a traditional and effective inference algorithm based on state space models (SSM), have been extensively applied in the fields of navigation and mapping. However, their performance will degrade when facing model assumption mismatches, such as non-linear dynamics and non-Gaussian correlated noises. The model-based deep learning methods overcome these mismatches by combining the domain knowledge of the model-based methods and the expressiveness of the data-driven deep learning methods, and thus can provide a promising solution for addressing high-dimensional and nonlinear challenges. This paper presents a succinct overview of the principles, inference model, and training methodology employed in model-based deep learning methods, with particular focus on the KalmanNet and the Dynamical Variational Autoencoder (DVAE). Furthermore, it implements KalmanNet on robust and high-precision navigation and positioning problem. The experimental results substantiate the feasibility of achieving navigation and positioning accuracy comparable to that of the Extended Kalman Filter (EKF), while simultaneously exhibiting enhanced robustness, albeit at the cost of some computational overhead.
- 13 December 2023
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-1/W, 1141-1146
Open AccessDYNAMICAL VARIATIONAL AUTOENCODERS AND KALMANNET: NEW APPROACHES TO ROBUST HIGH-PRECISION NAVIGATION
Kalman filters, recognized as a traditional and effective inference algorithm based on state space models (SSM), have been extensively applied in the fields of navigation and mapping. However, their performance will degrade when facing model assumption mismatches, such as non-linear dynamics and non-Gaussian correlated noises. The model-based deep learning methods overcome these mismatches by combining the domain knowledge of the model-based methods and the expressiveness of the data-driven deep learning methods, and thus can provide a promising solution for addressing high-dimensional and nonlinear challenges. This paper presents a succinct overview of the principles, inference model, and training methodology employed in model-based deep learning methods, with particular focus on the KalmanNet and the Dynamical Variational Autoencoder (DVAE). Furthermore, it implements KalmanNet on robust and high-precision navigation and positioning problem. The experimental results substantiate the feasibility of achieving navigation and positioning accuracy comparable to that of the Extended Kalman Filter (EKF), while simultaneously exhibiting enhanced robustness, albeit at the cost of some computational overhead.
- 22 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (7) , 12408-12418
ORB-NeuroSLAM: A Brain-Inspired 3-D SLAM System Based on ORB Features
Intelligent navigation is a fundamental technology that enables unmanned systems to achieve autonomy in the intelligent era. However, existing navigation schemes suffer from high computational complexity and power consumption, as well as low robustness in complex or unknown environments. To address these challenges, this article proposes a novel 3-D brain-inspired simultaneous localization and mapping (SLAM) method, called oriented FAST and rotated BRIEF (ORB)-NeuroSLAM, based on the ORB features. The proposed method takes inspiration from the robust and low-power navigation capabilities of humans and animals. The ORB-NeuroSLAM leverages the ORB features of camera images to compute robot self-motion and visual cues. Then, continuous attractor neural networks (CANNs) model multilayered head-direction cells and 3-D grid cells that exist in animal brains. These cells are utilized jointly to represent the robot poses. Efficient and robust methods for loop closure detection and experience map construction were also developed. The proposed method was verified on ten KITTI data sets, and experimental results demonstrate that it outperforms state-of-the-art brain-inspired SLAM methods in terms of accuracy and efficiency. Additionally, it is comparable to the state-of-the-art visual SLAM method ORB-SLAM3.
- 22 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (7) , 12408-12418
ORB-NeuroSLAM: A Brain-Inspired 3-D SLAM System Based on ORB Features
Intelligent navigation is a fundamental technology that enables unmanned systems to achieve autonomy in the intelligent era. However, existing navigation schemes suffer from high computational complexity and power consumption, as well as low robustness in complex or unknown environments. To address these challenges, this article proposes a novel 3-D brain-inspired simultaneous localization and mapping (SLAM) method, called oriented FAST and rotated BRIEF (ORB)-NeuroSLAM, based on the ORB features. The proposed method takes inspiration from the robust and low-power navigation capabilities of humans and animals. The ORB-NeuroSLAM leverages the ORB features of camera images to compute robot self-motion and visual cues. Then, continuous attractor neural networks (CANNs) model multilayered head-direction cells and 3-D grid cells that exist in animal brains. These cells are utilized jointly to represent the robot poses. Efficient and robust methods for loop closure detection and experience map construction were also developed. The proposed method was verified on ten KITTI data sets, and experimental results demonstrate that it outperforms state-of-the-art brain-inspired SLAM methods in terms of accuracy and efficiency. Additionally, it is comparable to the state-of-the-art visual SLAM method ORB-SLAM3.
- 31 January 2023
- Published by Springer Nature
- p. 3808-3816
A Brain-Inspired 3D Cognitive Map Construction Method Incorporating Monocular Visual Features
- 31 January 2023
- Published by Springer Nature
- p. 3808-3816
A Brain-Inspired 3D Cognitive Map Construction Method Incorporating Monocular Visual Features
- 23 May 2022
- Published by IOP Publishing in Metrologia
- Vol. 59 (3) , 035009
Research on a time scale algorithm based on multi-dimensional weighted average
The widespread availability of high-performance clocks has motivated interest in time scale algorithms. There are various time scale algorithms in use today, ranging in application from scientific to commercial. The weighted average algorithm has been widely used to generate an ensemble time scale. It takes one weight for each clock according to the stability or predictability to generate the ensemble time scale. However, one weight cannot reflect the long-term, mid-term and short-term stability performance of the clock simultaneously. Therefore, it cannot improve the long-term, mid-term and short-term stability at the same time. To solve this problem, a multi-dimensional weighted average algorithm is proposed in this paper, which decomposes the clock difference into three dimensions and gets weights in three dimensions to give consideration to improve long-term, mid-term and short-term stability simultaneously. The final results of simulation and experiment demonstrate that the algorithm proposed in this paper improved Allan deviations of 3.94 × 10−15 on daily and 7.86 × 10−15 on monthly averaging times with respect to those obtained from AT1 and ALGOS, respectively.
- 23 May 2022
- Published by IOP Publishing in Metrologia
- Vol. 59 (3) , 035009
Research on a time scale algorithm based on multi-dimensional weighted average
The widespread availability of high-performance clocks has motivated interest in time scale algorithms. There are various time scale algorithms in use today, ranging in application from scientific to commercial. The weighted average algorithm has been widely used to generate an ensemble time scale. It takes one weight for each clock according to the stability or predictability to generate the ensemble time scale. However, one weight cannot reflect the long-term, mid-term and short-term stability performance of the clock simultaneously. Therefore, it cannot improve the long-term, mid-term and short-term stability at the same time. To solve this problem, a multi-dimensional weighted average algorithm is proposed in this paper, which decomposes the clock difference into three dimensions and gets weights in three dimensions to give consideration to improve long-term, mid-term and short-term stability simultaneously. The final results of simulation and experiment demonstrate that the algorithm proposed in this paper improved Allan deviations of 3.94 × 10−15 on daily and 7.86 × 10−15 on monthly averaging times with respect to those obtained from AT1 and ALGOS, respectively.
- 8 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-8
Improving the Ensemble Time Scale by Eliminating Frequency Drifts of Hydrogen Masers
The long term stability of the time scale is affected by the large frequency drifts of hydrogen masers. It is necessary to calibrate the hydrogen maser’s drift before it is used to keep time. To calibrate drift, a reference with negligible drift is needed. Traceability links are established in timekeeping laboratories to get clock differences between local time scales and remote time scales. Drifts of remote time scales are negligible. Therefore, remote time scales can be used as reference for drift calibration. In this paper, the concept of remote virtual clock is proposed and the model is constructed. The remote virtual clock is involved in the local clock ensemble on timekeeping. With the help of the negligible frequency drift of virtual clock, drifts of local clocks can be calibrated. Meanwhile, a time scale algorithm based on drift calibration is proposed to generate the ensemble time scale. The final results of simulation and experiment demonstrate that the frequency drift can be calibrated and long term stability can be improved by one order of magnitude.
- 8 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-8
Improving the Ensemble Time Scale by Eliminating Frequency Drifts of Hydrogen Masers
The long term stability of the time scale is affected by the large frequency drifts of hydrogen masers. It is necessary to calibrate the hydrogen maser’s drift before it is used to keep time. To calibrate drift, a reference with negligible drift is needed. Traceability links are established in timekeeping laboratories to get clock differences between local time scales and remote time scales. Drifts of remote time scales are negligible. Therefore, remote time scales can be used as reference for drift calibration. In this paper, the concept of remote virtual clock is proposed and the model is constructed. The remote virtual clock is involved in the local clock ensemble on timekeeping. With the help of the negligible frequency drift of virtual clock, drifts of local clocks can be calibrated. Meanwhile, a time scale algorithm based on drift calibration is proposed to generate the ensemble time scale. The final results of simulation and experiment demonstrate that the frequency drift can be calibrated and long term stability can be improved by one order of magnitude.
- 1 August 2020
- Published by Canadian Science Publishing in Genome
- Vol. 63 (8) , 375-386
Evolution and domestication of Tc1/mariner transposons in the genome of African coelacanth (Latimeria chalumnae)
Here, we comprehensively analysed the abundance, diversity, and activity of Tc1/mariner transposons in African coelacanth (Latimeria chalumnae). Fifteen Tc1/mariner autonomous transposons were identified and grouped into six clades: DD34E/Tc1, DD34D/mariner, DD35D/Fot, DD31D/pogo, DD30-31D/pogo-like, and DD32–36D/Tigger, belonging to three known families: DD34E/Tc1, DD34D/mariner, and DD×D/pogo (DD35D/Fot, DD31D/pogo, DD30-31D/pogo-like, and DD32-36D/Tigger). Thirty-one miniature inverted-repeat transposable element (MITE) transposons of Tc1/mariner were also identified, and 20 of them display similarity to the identified autonomous transposons. The structural organization of these full Tc1/mariner elements includes a transposase gene flanked by terminal inverted repeats (TIRs) with TA dinucleotides. The transposases contain N-terminal DNA binding domain and a C-terminal catalytic domain characterized by the presence of a conservative D(Asp)DE(Glu)/D triad that is essential for transposase activity. The Tc1/mariner superfamily in coelacanth exhibited very low genome coverage (0.3%), but it experienced an extraordinary difference of proliferation dynamics among the six clades identified; moreover, most of them exhibited a very recent and current proliferation, suggesting that some copies of these transposons are putatively active. Additionally, at least four functional genes derived from Tc1/mariner transposons were found. We provide an up-to-date overview of Tc1/mariner in coelacanth, which may be helpful in determining genome and gene evolution in this living fossil.
- 1 August 2020
- Published by Canadian Science Publishing in Genome
- Vol. 63 (8) , 375-386
Evolution and domestication of Tc1/mariner transposons in the genome of African coelacanth (Latimeria chalumnae)
Here, we comprehensively analysed the abundance, diversity, and activity of Tc1/mariner transposons in African coelacanth (Latimeria chalumnae). Fifteen Tc1/mariner autonomous transposons were identified and grouped into six clades: DD34E/Tc1, DD34D/mariner, DD35D/Fot, DD31D/pogo, DD30-31D/pogo-like, and DD32–36D/Tigger, belonging to three known families: DD34E/Tc1, DD34D/mariner, and DD×D/pogo (DD35D/Fot, DD31D/pogo, DD30-31D/pogo-like, and DD32-36D/Tigger). Thirty-one miniature inverted-repeat transposable element (MITE) transposons of Tc1/mariner were also identified, and 20 of them display similarity to the identified autonomous transposons. The structural organization of these full Tc1/mariner elements includes a transposase gene flanked by terminal inverted repeats (TIRs) with TA dinucleotides. The transposases contain N-terminal DNA binding domain and a C-terminal catalytic domain characterized by the presence of a conservative D(Asp)DE(Glu)/D triad that is essential for transposase activity. The Tc1/mariner superfamily in coelacanth exhibited very low genome coverage (0.3%), but it experienced an extraordinary difference of proliferation dynamics among the six clades identified; moreover, most of them exhibited a very recent and current proliferation, suggesting that some copies of these transposons are putatively active. Additionally, at least four functional genes derived from Tc1/mariner transposons were found. We provide an up-to-date overview of Tc1/mariner in coelacanth, which may be helpful in determining genome and gene evolution in this living fossil.