Article
Computer Science, Artificial Intelligence
Jifeng Guo, Zhiqi Pang, Miaoyuan Bai, Yanbang Xiao, Jian Zhang
Summary: The core idea of active learning is to achieve higher model performance at a reduced annotation cost. This paper introduces an independency-enhancing adversarial active learning method, which differs from previous approaches by emphasizing sample independence. The informativeness of a group of samples is believed to be related to sample independence rather than the simple sum of individual sample informativeness. To ensure sample independence, an independent sample selection module based on hierarchical clustering is designed. An adversarial approach is also utilized to learn the feature representation and label the state of the sample based on predicted loss value. Sample selection is performed based on sample uncertainty, diversity, and independence. Experimental results on four datasets demonstrate the effectiveness and superiority of this independency-enhancing adversarial active learning approach.
IET IMAGE PROCESSING
(2023)
Article
Chemistry, Analytical
Xianyi Chen, Xiafu Peng, Sun'an Wang
Summary: This paper proposes a method using density peak clustering to reduce redundant superpixels and highlight the primary textures and contours of salient objects. Experimental results show good performance in efficiency and accuracy.
Article
Computer Science, Artificial Intelligence
Daquan Zhou, Qibin Hou, Linjie Yang, Xiaojie Jin, Jiashi Feng
Summary: This article explores the token selection behavior of self-attention and proposes simple approaches to enhance selectivity and diversity. By developing a token selector module, it significantly boosts the performance of various ViT backbones. These approaches allow ViTs to achieve high accuracy with a relatively small number of parameters and can be applied to different models for image classification, semantic segmentation, and NLP tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Bernat Galmes, Gabriel Moya-Alcover, Pedro Bibiloni, Javier Varona, Antoni Jaume-i-Capo
Summary: This article presents a robust segmentation method for measuring toenails. The method is used in a clinical trial to objectively quantify the incidence of a specific pathology. It uses the Hough transform to locate the tip of the toe and estimate the nail location and size, and then classifies the super-pixels based on their geometric and photometric information. The watershed transform is then used to delineate the border of the nail.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Aram Ter-Sarkisov
Summary: A novel framework that integrates lesion mask segmentation and COVID-19 prediction achieved high sensitivity and accuracy on the test split of the CNCB-NCOV dataset by introducing the concept of affinities for classifying the whole input image.
APPLIED SOFT COMPUTING
(2022)
Article
Environmental Sciences
Alexandru Pop, Victor Domsa, Levente Tamas
Summary: In this paper, a novel rotation normalization technique using an oriented bounding box for point cloud processing is proposed. It is used to create a point cloud annotation tool for part segmentation and trained on custom datasets for classification and part segmentation tasks. The method is successfully deployed on an embedded device with limited processing power and compared with other rotation-invariant features in noisy synthetic datasets. Our method offers more auxiliary information related to the object's dimension, position, and orientation while performing at a similar level.
Article
Ecology
Jarrett Blair, Michael D. Weiser, Kirsten de Beurs, Michael Kaspari, Cameron Siler, Katie E. Marshall
Summary: This study presents a practical methodology of using machine learning in ecological data acquisition pipelines, training and testing algorithms to classify a large number of terrestrial invertebrate specimens. The study addresses issues of inconsistent taxonomic label specificity and unknown taxa classification. The results show that complex machine learning methods are not necessarily more accurate than traditional methods, and the inclusion of contextual metadata improves accuracy.
ECOLOGICAL INFORMATICS
(2022)
Article
Computer Science, Information Systems
Hang Cheng, Shugong Xu, Fengjun Guo
Summary: This paper proposes a matting method that utilizes flexible interactive guidance including trimap, scribblemap, clickmap, and bounding box. The proposed method gradually transforms trimaps into other guidance inputs during training to maintain accuracy while simplifying input. Experimental results show that the method achieves competitive results compared with existing trimap-based and trimap-free methods.
Article
Radiology, Nuclear Medicine & Medical Imaging
Adam Berkley, Camillo Saueressig, Utkarsh Shukla, Imran Chowdhury, Anthony Munoz-Gauna, Olalekan Shehu, Ritambhara Singh, Reshma Munbodh
Summary: State-of-the-art deep learning models show promising performance on cross-institutional predictions. They considerably improve on previous models and can transfer knowledge to new types of brain tumors without additional modeling.
Article
Computer Science, Artificial Intelligence
Lukasz Karbowiak, Janusz Bobulski
Summary: This article introduces the importance of background segmentation and proposes a method to compare algorithms under severe weather conditions. Through testing in different weather conditions, interesting differences in detail detection and detection noise were observed.
PEERJ COMPUTER SCIENCE
(2022)
Article
Biotechnology & Applied Microbiology
Xinyi Shen, Guolong Shi, Huan Ren, Wu Zhang
Summary: This paper proposes an improved YOLO algorithm with an increased vertical grid number, which enables real-time detection of targets in high-resolution zoom sensing images. By extracting the light and shade levels and feature parameters from the grey-level cooccurrence matrix, and using the SLIC superpixel segmentation method for scene segmentation, the algorithm achieves higher accuracy and real-time performance compared to the original YOLO algorithm.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Julian Luengo, Raul Moreno, Ivan Sevillano, David Charte, Adrian Pelaez-Vegas, Marta Fernandez-Moreno, Pablo Mesejo, Francisco Herrera
Summary: This paper reviews and categorizes computer vision techniques for metallographic image segmentation, introduces deep learning-based ensemble techniques utilizing pixel similarity, and conducts thorough comparisons in real-world datasets to discuss strengths, weaknesses, and application frameworks. The paper also addresses open challenges in the field to provide guidance for future research to fill existing gaps.
INFORMATION FUSION
(2022)
Article
Energy & Fuels
Robinson Cavieres, Rodrigo Barraza, Danilo Estay, Jose Bilbao, Patricio Valdivia-Lefort
Summary: This article introduces an artificial neural network tool to quantify power loss in solar photovoltaic modules due to soiling and partial shading effects. The proposed method consists of three main stages: segmentation, resizing, and performance prediction. Compared to state-of-the-art computer vision architectures, the approach achieves similar results with a significant reduction in computational cost.
Review
Computer Science, Artificial Intelligence
Yuelong Chuang, Shiqing Zhang, Xiaoming Zhao
Summary: In recent years, there has been increasing attention on panoptic segmentation, resulting in the emergence of numerous related algorithms. Deep neural networks have been commonly employed in panoptic segmentation due to the success of deep learning methods in other tasks. This article provides a comprehensive exploration of panoptic segmentation, with a focus on analyzing and understanding RGB image data. The authors discuss the background of panoptic segmentation, including deep learning models and image segmentation, and cover various related topics such as datasets, evaluation metrics, models, and subfields. They also examine the challenges and future directions in this area.
IET IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Yan Zhou, Xihong Zheng, Yin Yang, Jianxun Li, Jinzhen Mu, Richard Irampaye
Summary: This study proposes a network named MRFNet based on two-branch strategy for efficient and accurate semantic segmentation in urban scenes. The network utilizes a Multi-directional Feature Refinement Module (MFRM) to comprehensively consider contextual information from sub-regions in different directions and at different scales. The network also introduces a Feature Cross-guide Aggregation Module to aggregate detailed information and contextual information through mutual guidance.
IET COMPUTER VISION
(2023)
Article
Automation & Control Systems
Rahul Kumar, Uday Pratap Singh, Arun Bali, Kuldip Raj
Summary: An adaptive hybrid neural control scheme is proposed for uncertain non-linear discrete-time systems with non-symmetric dead-zone input and unknown disturbances. The scheme overcomes the complexity of controlling such systems by introducing an adaptive compensative term for the non-symmetric dead-zone and constructing a hybrid neural network controller. Simulation examples validate the effectiveness of the proposed scheme, including an example inspired by a real-world system called continuous stirred tank reactor.
INTERNATIONAL JOURNAL OF CONTROL
(2023)
Review
Pathology
Neha Kumari, Ritu Verma, Vinita Agrawal, Uday Pratap Singh
Summary: This study retrospectively analyzed the clinical features, morphological characteristics, and outcome of renal neuroendocrine tumors. Six patients who underwent radical nephrectomy were included in the study. The results showed that these tumors were mostly well-differentiated and had a low risk of metastasis or relapse.
INTERNATIONAL JOURNAL OF SURGICAL PATHOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Arun Bali, Uday Pratap Singh, Rahul Kumar
Summary: This paper investigates the problem of Multi-dimensional Taylor Network (MTN)-based fault-tolerant control (FTC) for single-input and single-output nonlinear systems in non-strict feedback form. A MTN-based FTC method is presented for nonlinear systems with actuator faults and unmodeled dynamics. The proposed technique ensures that all closed-loop system signals are semi-globally uniformly ultimately bounded (SGUUB) and the tracking error converges to a small region around the origin. Three examples, including a single-link robot manipulator, are presented to demonstrate the effectiveness of the proposed controller design.
NEURAL PROCESSING LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Arun Bali, Uday Pratap Singh, Rahul Kumar
Summary: This work addresses the problem of adaptive neural control for nonlinear systems with input nonlinearities and sensor fault in non-strict feedback form. Radial basis function neural networks are used to approximate the unknown functions in the system, and an adaptive neural controller is developed using the backstepping method. The proposed control method ensures the stability of the system and the tracking performance of the output based on Lyapunov stability theory.
Article
Computer Science, Information Systems
Jasvinder Pal Singh, Uday Pratap Singh, Sanjeev Jain
Summary: This paper focuses on recognizing people in a multi-gait scenario. It presents a model-based approach to reconstruct occluded regions and extract linear kinematic features. A hybrid classifier is proposed for multi-gait identification and experimental results demonstrate its superiority over existing methods.
MULTIMEDIA SYSTEMS
(2023)
Article
Computer Science, Information Systems
Shubhangi Solanki, Uday Pratap Singh, Siddharth Singh Chouhan, Sanjeev Jain
Summary: A brain tumor is caused by rapid and uncontrolled cell growth, which can be fatal if not treated early. Despite significant efforts, accurate segmentation and classification of brain tumors remain challenging due to variations in location, structure, and proportions. This study aims to provide researchers with comprehensive literature on the ability of Magnetic Resonance imaging to identify brain tumors. Using computational intelligence and statistical image processing techniques, the research proposes several methods for detecting brain cancer and tumors. The study also includes an evaluation matrix for specific systems and dataset types, as well as discussions on tumor morphology, available datasets, augmentation methods, and categorization of Deep Learning, Transfer Learning, and Machine Learning models. Additionally, the research compiles relevant information on tumor identification, including benefits, drawbacks, advancements, and future trends.
Article
Engineering, Electrical & Electronic
Arun Bali, Siddharth Singh Chouhan, Gourav Kumar, Rahul Kumar, Uday Pratap Singh
Summary: In this article, the problem of adaptive fault-tolerant control for a class of stochastic pure-feedback nonlinear systems with simultaneous actuator and sensor faults is examined. The stochastic pure-feedback nonlinear system is converted into a strict-feedback system using the mean value theorem, and unknown functions are approximated using radial basis function neural networks. By determining the greatest value of the norm of the neural network weight vector, only one adaptive parameter needs to be calculated online. The unavailability of state variables caused by sensor faults is addressed using regrouping and parameter separation methods. The Lyapunov function methods and backstepping recursive design technique are used to design an adaptive fault-tolerant controller, which ensures convergence of tracking errors and boundedness of all signals in the closed-loop system.
CIRCUITS SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Automation & Control Systems
Arun Bali, Uday Pratap Singh, Rahul Kumar, Sanjeev Jain
Summary: An adaptive finite-time fault-tolerant control technique is proposed for a class of switched nonlinear systems in strict-feedback form. The system is affected by actuator fault, input dead-zone, and external disturbances. Radial basis function neural networks (RBFNNs) are used to approximate unknown functions and minimize the negative effect of faults. The proposed control method, based on neural networks and the backstepping method, ensures both transient and steady-state control performance and has been shown to be effective in simulation examples, including a continuous stirred tank reactor (CSTR) application. (c) 2023 European Control Association. Published by Elsevier Ltd. All rights reserved.
EUROPEAN JOURNAL OF CONTROL
(2023)
Article
Computer Science, Information Systems
Misbah Shafi, Rakesh Kumar Jha, Sanjeev Jain
Summary: The advancement in wireless communication technologies necessitates the need for improved security measures, particularly in detecting attacks such as spoofing and signal strength attacks. This paper proposes an Intrusion Detection System based on graph theory to identify attacked nodes in the communication network. The algorithm analyzes the network layer by layer to extract vulnerable nodes and determine the attacked node(s). The IDS strategy is based on energy efficiency and secrecy rate analysis, detecting nodes with values beyond specified thresholds. The proposed approach outperforms conventional intrusion detection methods in terms of performance, computation time, and complexity.
IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT
(2023)
Article
Computer Science, Software Engineering
Gourav Kumar, Uday Pratap Singh, Sanjeev Jain
Summary: The stock price fluctuations of different countries are interrelated, and this study examines the interrelationship among Asian stock markets and forecast the stock market based on this relationship. The Granger causality (GC) test and Pearson's correlation (PC) matrix are used to test the interrelationship. The results show a strong correlation among Asian stock markets, and the GC-LSTM model outperforms the PC-LSTM model in forecasting performance.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Meeting Abstract
Urology & Nephrology
Sanjoy Kumar Sureka, Sumit Mandal, Uday Pratap Singh
JOURNAL OF UROLOGY
(2023)
Article
Computer Science, Information Systems
Sheikh Imroza Manzoor, Sanjeev Jain, Yashwant Singh, Harvinder Singh
Summary: Our daily lives are significantly influenced by intelligent IoT applications, services, devices, and industries. AI is expected to have a major impact on training machine learning algorithms on IoT devices without sharing data. Federated learning has emerged as a popular research area to address privacy concerns when training machine learning models across IoT devices. However, there are challenges such as privacy, effectiveness, and efficiency. This article proposes a taxonomy of FL-based IoT systems, analyzes related works, studies threats and attacks, devises privacy-preserving FL techniques, and highlights open research challenges.
Article
Urology & Nephrology
Sanjoy Kumar Sureka, Ankit Misra, Himanshu Raj, Anupam Shukla, Uday Pratap Singh
Summary: The efficacy of 2-core prostate biopsy in advanced prostate cancer patients was assessed. A retrospective analysis of 12-core prostate biopsies and a prospective validation were conducted to determine if a reduced number of cores are sufficient for histopathological diagnosis.
INTERNATIONAL UROLOGY AND NEPHROLOGY
(2023)
Article
Engineering, Multidisciplinary
Sicheng Jiao, Shixiang Wang, Minge Gao, Min Xu
Summary: This paper presents a non-contact method of thickness measurement for thin-walled rotary shell parts based on a chromatic confocal sensor. The method involves using a flip method to obtain surface profiles from both sides of the workpiece, measuring the decentration and tilt errors of the workpiece using a centering system, establishing a unified reference coordinate system, reconstructing the external and internal surface profiles, and calculating the thickness. Experimental results show that the method can accurately measure the thickness of a sapphire spherical shell workpiece and is consistent with measurements of other materials.
Article
Engineering, Multidisciplinary
Rajeev Kumar, Sajal Agarwal, Sarika Pal, Alka Verma, Yogendra Kumar Prajapati
Summary: This study evaluated the performance of a CaF2-Ag-MXene-based surface plasmon resonance (SPR) sensor at different wavelengths. The results showed that the sensor achieved the maximum sensitivity at a wavelength of 532 nm, and higher sensitivities were obtained at shorter wavelengths at the expense of detection accuracy.
Article
Engineering, Multidisciplinary
Attilio Di Nisio, Gregorio Andria, Francesco Adamo, Daniel Lotano, Filippo Attivissimo
Summary: Capacitive sensing is a widely used technique for a variety of applications, including avionics. However, current industry standard Capacitive Level Sensors (CLSs) used in helicopters perform poorly in terms of sensitivity and dynamic characteristics. In this study, novel geometries were explored and three prototypes were built and tested. Experimental validation showed that the new design featuring a helicoidal slit along the external electrode of the cylindrical probe improved sensitivity, response time, and linearity.
Article
Engineering, Multidisciplinary
Kai Yang, Huiqin Wang, Ke Wang, Fengchen Chen
Summary: This paper proposes an effective measurement method for dynamic compaction construction based on time series model, which enables real-time monitoring and measurement of anomalies and important construction parameters through simulating motion state transformation and running time estimation.
Article
Engineering, Multidisciplinary
Hui Fu, Qinghua Song, Jixiang Gong, Liping Jiang, Zhanqiang Liu, Qiang Luan, Hongsheng Wang
Summary: An automatic detection and pixel-level quantification model based on joint Mask R-CNN and TransUNet is developed to accurately evaluate microcrack damage on the grinding surfaces of engineering ceramics. The model is effectively trained on actual micrograph image dataset using a joint training strategy. The proposed model achieves reliable automatic detection and fine segmentation of microcracks, and a skeleton-based quantification model is also proposed to provide comprehensive and precise measurements of microcrack size.
Review
Engineering, Multidisciplinary
Sang Yeob Kim, Da Yun Kwon, Arum Jang, Young K. Ju, Jong-Sub Lee, Seungkwan Hong
Summary: This paper reviews the categorization and applications of UAV sensors in forensic engineering, with a focus on geotechnical, structural, and water infrastructure fields. It discusses the advantages and disadvantages of sensors with different wavelengths and addresses the challenges of current UAV technology and recommendations for further research in forensic engineering.
Article
Engineering, Multidisciplinary
Anton Nunez-Seoane, Joaquin Martinez-Sanchez, Erik Rua, Pedro Arias
Summary: This article compares the use of Mobile Laser Scanners (MLS) and Aerial Laser Scanners (ALS) for digitizing the road environment and detecting road slopes. The study found that ALS data and its corresponding algorithm achieved better detection and delimitation results compared to MLS. Measuring the road from a terrestrial perspective negatively impacted the detection process, while an aerial perspective allowed for scanning of the entire slope structure.
Article
Engineering, Multidisciplinary
Nur Luqman Saleh, Aduwati Sali, Raja Syamsul Azmir Raja Abdullah, Sharifah M. Syed Ahmad, Jiun Terng Liew, Fazirulhisyam Hashim, Fairuz Abdullah, Nur Emileen Abdul Rashid
Summary: This study introduces an enhanced signal processing scheme for detecting mouth-click signals used by blind individuals. By utilizing additional band-pass filtering and other steps, the detection accuracy is improved. Experimental results using artificial signal data showed a 100% success rate in detecting obstacles. The emerging concepts in this research are expected to benefit radar and sonar system applications.
Article
Engineering, Multidisciplinary
Jiqiang Tang, Shengjie Qiu, Lu Zhang, Jinji Sun, Xinxiu Zhou
Summary: This paper studies the magnetic noise level of a compact high-performance magnetically shielded room (MSR) under different operational conditions and establishes a quantitative model for magnetic noise calculation. Verification experiments show the effectiveness of the proposed method.
Review
Engineering, Multidisciplinary
Krzysztof Bartnik, Marcin Koba, Mateusz Smietana
Summary: The demand for miniaturized sensors in the biomedical industry is increasing, and optical fiber sensors (OFSs) are gaining popularity due to their small size, flexibility, and biocompatibility. This study reviews various OFS designs tested in vivo and identifies future perspectives and challenges for OFS technology development from a user perspective.
Article
Engineering, Multidisciplinary
Yue Wang, Lei Zhou, Zihao Li, Jun Wang, Xuangou Wu, Xiangjun Wang, Lei Hu
Summary: This paper presents a 3-D reconstruction method for dynamic stereo vision of metal surface based on line structured light, overcoming the limitation of the measurement range of static stereo vision. The proposed method uses joint calibration and global optimization to accurately reconstruct the 3-D coordinates of the line structured light fringe, improving the reconstruction accuracy.
Article
Engineering, Multidisciplinary
Jaafar Alsalaet
Summary: Order tracking analysis is an effective tool for machinery fault diagnosis and operational modal analysis. This study presents a new formulation for the data equation of the second-generation Vold-Kalman filter, using separated cosine and sine kernels to minimize error and provide smoother envelopes. The proposed method achieves high accuracy even with small weighting factors.
Article
Engineering, Multidisciplinary
Tonglei Cao, Kechen Song, Likun Xu, Hu Feng, Yunhui Yan, Jingbo Guo
Summary: This study constructs a high-resolution dataset for surface defects in ceramic tiles and addresses the scale and quantity differences in defect distribution. An improved approach is proposed by introducing a content-aware feature recombination method and a dynamic attention mechanism. Experimental results demonstrate the superior accuracy and efficiency of the proposed method.
Article
Engineering, Multidisciplinary
Qinghong Fu, Yunxi Lou, Jianghui Deng, Xin Qiu, Xianhua Chen
Summary: Measurement and quantitative characterization of aging-induced gradient properties is crucial for accurate analysis and design of asphalt pavement. This research proposes the composite specimen method to obtain asphalt binders at different depths within the mixture and uses dynamic shear rheometer tests to measure aging-induced gradient properties and reveal internal mechanisms. G* master curves are constructed to investigate gradient aging effects in a wide range. The study finds that the composite specimen method can effectively restore the boundary conditions and that it is feasible to study gradient aging characteristics within the asphalt mixture. The study also observes variations in G* and delta values and the depth range of gradient aging effects for different aging levels.
Article
Engineering, Multidisciplinary
Min Li, Kai Wei, Tianhe Xu, Yali Shi, Dixing Wang
Summary: Due to the limitations of ground monitoring stations in China for the BDS, the accuracy of BDS Medium Earth Orbit (MEO) satellite orbits can be influenced. To overcome this, low Earth orbit (LEO) satellites can be used as additional monitoring stations. In this study, data from two LEO satellites were collected to improve the precise orbit determination of the BDS. By comparing the results with GPS and BDS-2/3 solutions, it was found that including the LEO satellites significantly improved the accuracy of GPS and BDS-2/3 orbits.
