Article
Plant Sciences
Bing Liu, Luyang Liu, Ran Zhuo, Weidong Chen, Rui Duan, Guishen Wang
Summary: This paper presents a dataset for forestry pest identification, consisting of 31 categories of pests and their different forms. Experimental results show that the dataset performs well on various models. The researchers hope that this dataset will be helpful for pest control and pest detection research in the future.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Agronomy
Maria Teresa Linaza, Jorge Posada, Jurgen Bund, Peter Eisert, Marco Quartulli, Juergen Doellner, Alain Pagani, Igor G. Olaizola, Andre Barriguinha, Theocharis Moysiadis, Laurent Lucat
Summary: One of the main challenges for implementing artificial intelligence in agriculture is its low replicability and difficulty in systematic data gathering due to different field conditions. However, by comparing pilot experiments in various fields, collective knowledge can be enhanced. Current research activities focus on European countries, aiming to present achieved results, ongoing investigations, and technical challenges.
Article
Engineering, Mechanical
Shanwu Li, Yongchao Yang
Summary: A new data-driven framework based on physics-integrated deep learning is proposed for nonlinear modal identification of unknown nonlinear dynamical systems solely from response data. The method effectively identifies NNMs with invariant manifolds, energy-dependent nonlinear modal spectrum, and future-state prediction, consistent with results from theoretical derivation or numerical computation.
NONLINEAR DYNAMICS
(2021)
Article
Entomology
Jia-Hsin Huang, Yu-Ting Liu, Hung Chih Ni, Bo-Ye Chen, Shih-Ying Huang, Huai-Kuang Tsai, Hou-Feng Li
Summary: This study developed an automated deep learning classifier for termite image recognition using MobileNetV2 model, achieving high accuracy in identifying different castes of termite pests. Image augmentation techniques were applied to reduce the number of original images required for classification without sacrificing accuracy.
JOURNAL OF ECONOMIC ENTOMOLOGY
(2021)
Article
Computer Science, Information Systems
Makoto Ikeda, Natwadee Ruedeeniraman, Leonard Barolli
Summary: The VegeCareAI agricultural support system is proposed to assist agricultural workers with vegetable classification, plant disease classification, and insect pest classification to improve crop productivity. Experimental results show the system has advantages in supporting various crops, but faces challenges in insect pest classification accuracy for different life cycles.
INTERNET OF THINGS
(2021)
Article
Biology
Xiaoyu Wang, Nitin Patil, Fuyi Li, Zhikang Wang, Haolan Zhan, Daniel Schmidt, Philip Thompson, Yuming Guo, Cornelia B. Landersdorfer, Hsin-Hui Shen, Anton Y. Peleg, Jian Li, Jiangning Song
Summary: In this study, we developed a machine learning framework called PmxPred for predicting polymyxin analogues with high antimicrobial activity against Gram-negative bacteria. The framework achieved good performance on multiple datasets, outperforming traditional transfer learning methods.
COMPUTERS IN BIOLOGY AND MEDICINE
(2024)
Article
Energy & Fuels
Imen Jendoubi, Francois Bouffard
Summary: With the ongoing energy transition, electric power and energy systems are becoming increasingly multi-dimensional and complex with higher levels of uncertainty. Reinforcement learning provides a data-driven alternative to traditional model-based control methods, allowing efficient control of such systems without prior system dynamics modeling or predictions. This study proposes a multi-agent deep reinforcement learning-based control framework for solving multi-dimensional power dispatch problems in systems with multiple uncertainties.
SUSTAINABLE ENERGY GRIDS & NETWORKS
(2022)
Article
Plant Sciences
Satinder Bal Gupta, RajKumar Yadav, Fatemeh Bovand, Pankaj Kumar Tyagi
Summary: Castor is an important nonedible industrial crop that produces oil and is used in various products. Insect pest attacks can degrade the quality and quantity of castor oil. Traditional methods of identifying pests are time-consuming and require expertise. This study proposes a hybrid data augmentation approach to solve the lack of a suitable dataset for effective vision-based model training, using deep convolutional neural networks. The results show that this method improves overall performance compared to previous methods.
FRONTIERS IN PLANT SCIENCE
(2023)
Article
Automation & Control Systems
Daixin Fu, Lingyi Wang, Guanlin Lv, Zhengyu Shen, Hao Zhu, W. D. Zhu
Summary: This paper provides a comprehensive review of dynamic load identification methods based on data-driven techniques. It covers two aspects: load localization and load reconstruction, and discusses various data-driven techniques. Additionally, the paper offers insight into the challenges and prospects of the data-driven techniques for dynamic load identification.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Justin Engelmann, Alice D. McTrusty, Ian J. C. MacCormick, Emma Pead, Amos Storkey, Miguel O. Bernabeu
Summary: This study proposes an improved deep learning model for recognizing multiple retinal diseases under more realistic conditions and uses global explainability methods to identify the relevant regions in ultra-widefield (UWF) images. The model performs well in distinguishing between healthy and diseased retinas and identifies the posterior pole as the most important region.
NATURE MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Hardware & Architecture
Jiachen Yang, Guipeng Lan, Yang Li, Yicheng Gong, Zhuo Zhang, Sezai Ercisli
Summary: This paper investigates the issue of data quality constraint on deep learning models in the field of smart agriculture and proposes two methods for assessing data quality. Experimental results demonstrate that the used public dataset has a large amount of redundancy, and the rules for selecting informative samples are significant for the design of efficient datasets.
COMPUTERS & ELECTRICAL ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Jiachen Yang, Zhuo Zhang, Shuai Xiao, Shukun Ma, Yang Li, Wen Lu, Xinbo Gao
Summary: With the development of computer vision, the research on human activity understanding has been greatly promoted. This paper proposes a core weight entropy data information evaluation method based on feature distribution analysis, which effectively reduces data consumption and achieves high performance using a small amount of high information human activity data.
Article
Agriculture, Multidisciplinary
Offer Rozenstein, Yafit Cohen, Victor Alchanatis, Karl Behrendt, David J. Bonfil, Gil Eshel, Ally Harari, W. Edwin Harris, Iftach Klapp, Yael Laor, Raphael Linker, Tarin Paz-Kagan, Sven Peets, S. Mark Rutter, Yael Salzer, James Lowenberg-DeBoer
Summary: Sustainability in food and fiber agriculture systems relies on knowledge, technology, and data-driven approaches rather than solely on human observation and experience. Data collected by sensors and analyzed by artificial intelligence can help farmers understand the interactions between natural systems to achieve both sustainability and food security.
PRECISION AGRICULTURE
(2023)
Article
Agriculture, Multidisciplinary
Sahil Verma, Prabhat Kumar, Jyoti Prakash Singh
Summary: Plant diseases pose a significant threat to global food security and economic prosperity. Deep learning models based on Convolution Neural Network (CNN) have shown promise in addressing plant disease detection tasks. This study proposes a meta-learning-based framework that recommends top-n suitable models for unseen plant disease detection datasets using prior evaluations of benchmark models.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Green & Sustainable Science & Technology
Chen-Yu Tai, Wun-Jhe Wang, Yueh-Min Huang
Summary: Adequate data is essential for accurate trend prediction. This study utilizes TimeGAN to generate agricultural sensing data and train neural network models to predict future pest populations. The results show that the generated data effectively compensates for the lack of actual data and produces similar predictions to those trained on actual data. Accurate prediction of pest populations would be a breakthrough in pest control.
Article
Automation & Control Systems
Yang Li, Jiachen Yang, Zhuo Zhang, Jiabao Wen, Prabhat Kumar
Summary: This article proposes a normalized double entropy (NDE) method to assess image data quality, and the experimental results demonstrate its stability and effectiveness. Comparisons driven by selected good and bad data show that even with 70% of the dataset, the accuracy can be maintained. This is of great significance for improving efficiency and cybersecurity in healthcare systems.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Electrical & Electronic
Yang Zhao, Shuai Xiao, Jiachen Yang, Wen Lu, Xinbo Gao
Summary: This paper introduces a method for accurately predicting the quality of TMIs: RETI. Based on the characteristics of HDR images, three important elements including authenticity, energy and information preservation, and scene expressiveness are considered, combined with subjective quality for training. The results show that the method has good prediction and generalization abilities compared to some state-of-the-art methods.
Article
Engineering, Electrical & Electronic
Chengang Lyu, Xiaojiao Lin, Mengqi Zhang, Chunfeng Ge, Jiachen Yang
Summary: In the unattended industrial environment, the combination of image detection and machine learning for piping leakage detection algorithm provides high sensitivity and reliability. However, practical application faces challenges such as the need to detect a large number of industrial leakage scenes, significant spatial feature differences between deployed and undeployed scenes, and unpredictable actual leakage situations. To address these challenges, a piping leakage detection method with domain generalization (DG) capability is proposed, utilizing AMSRCR for spatial feature enhancement and designing DCM-DenseNet121 for improved model generalization. Experimental results demonstrate a leakage detection balanced accuracy (BA) of 99.7% when inferring on three target scenes, with efficient detection in extra-domain target scenes while maintaining relatively lightweight and meeting industrial field bandwidth limitations.
IEEE SENSORS JOURNAL
(2023)
Article
Engineering, Civil
Judy P. Yang, Ching-Chien Wang
Summary: This study aims to derive analytical formulations for one and two amplifier-enhanced vehicle-bridge systems, and verify their correctness through dynamic responses and finite element formulations. The results show good agreement between the analytical formulations and the finite element methods, and clear amplifier spectra of scanned bridge frequencies.
INTERNATIONAL JOURNAL OF STRUCTURAL STABILITY AND DYNAMICS
(2023)
Article
Computer Science, Information Systems
Zhengjian Li, Jiabao Wen, Jiachen Yang, Jingyi He, Tianlei Ni, Yang Li
Summary: This article proposes a scheme of energy-efficient autonomous and decentralized sensor network called SAGOI-Net, using an intelligent autonomous underwater glider (AUG) to minimize energy consumption. The proposed network utilizes a nonpropeller-driven AUG without acoustic or vision sensors, and equips a lightweight neural network-based self-navigation system to save energy consumption. Training a lightweight SN-LSTM model, the impact of the ocean environment on AUG is quantitatively analyzed, and simulation results show the superior performance of the AUG SN and energy efficiency of the proposed SAGOI-Net.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Automation & Control Systems
Jiachen Yang, Jingfei Ni, Meng Xi, Jiabao Wen, Yang Li
Summary: This study proposes a path planning algorithm for underwater robots that uses a three-dimensional marine environment and an experience screening mechanism to achieve obstacle avoidance. The algorithm is validated to have better performance compared to traditional methods, providing improved autonomy and stability for underwater robots in dynamic environments.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Chemistry, Analytical
Siyuan Hu, Shuai Xiao, Jiachen Yang, Zuochen Zhang, Kunyu Zhang, Yong Zhu, Yubo Zhang
Summary: This paper proposes a model called N-DDQNP to address the static path planning problem in ocean environments. The N-DDQNP model combines a noise network and a prioritized experience replay mechanism to improve the exploration and convergence speed issues of the DQN algorithm. Experiments using real ocean data demonstrate that the N-DDQNP model outperforms other algorithms in various ocean current scenarios and obstacle environments.
Article
Automation & Control Systems
Jiachen Yang, Yanshuang Zhou, Yang Zhao, Wen Lu, Xinbo Gao
Summary: This paper introduces a metalearning-based multipatch IAA method that can quickly adapt to various thematic tasks and proposes a complete-information patch selection scheme and multipatch network. Experimental results demonstrate the superiority of this method and provide valuable guidance for network initialization of IAA.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Engineering, Multidisciplinary
Jiabao Wen, Jiachen Yang, Yang Li, Jingyi He, Zhengjian Li, Houbing Song
Summary: The new generation of artificial intelligence technology has enhanced the autonomous monitoring capabilities of marine equipment. A ocean monitoring platform based on edge computing enables autonomous collaboration among multiple equipment groups. By using an improved artificial potential field method scheme, the challenges faced by multi-AUG systems operating in special underwater environments can be overcome, enabling cooperative control of the AUG group.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2023)
Article
Telecommunications
Jiabao Wen, Jiachen Yang, Tianying Wang, Yang Li, Zhihan Lv
Summary: In order to efficiently complete a complex computation task, it is necessary to decompose the task into subcomputation tasks that run parallel in edge computing. Wireless Sensor Network (WSN) is an example of parallel computation. A task allocation strategy is needed to reduce energy consumption and balance the load of the network, which is crucial for achieving highly reliable parallel computation in WSN.
DIGITAL COMMUNICATIONS AND NETWORKS
(2023)
Article
Computer Science, Artificial Intelligence
Meng Xi, Jiachen Yang, Jiabao Wen, Zhengjian Li, Wen Lu, Xinbo Gao
Summary: An autonomous underwater vehicle (AUV) has shown potential in marine missions, with path planning as an essential prerequisite. However, existing methods face limitations due to the gap with reality and the demands of dynamic and unknown environments. To address these limitations, we propose an information-assisted reinforcement learning path planning scheme, which includes establishing a realistic simulation environment based on ocean current observations, using information compression to improve generalization, and designing a confidence evaluator for action in dynamic environments.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Telecommunications
Yang Li, Jiachen Yang, Jiabao Wen
Summary: The ongoing data explosion has posed unprecedented challenges to the information security of communication networks. This study focuses on the data redundancy analysis and screening of images, which are commonly used as information transmission carriers. Through extensive experiments, it is found that open datasets from different domains exhibit serious data redundancy. A novel entropy-based information screening method is proposed, outperforming random sampling under various experimental conditions.
DIGITAL COMMUNICATIONS AND NETWORKS
(2023)
Article
Computer Science, Artificial Intelligence
Jiachen Yang, Jipeng Zhang
Summary: This article discusses the vulnerability of deep reinforcement learning algorithms and proposes a solution called robust adversary populations with volume diversity measure (RAP Vol) to improve their robustness. The proposed method updates all adversaries and enhances population diversity to improve generalization performance.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
