Article
Computer Science, Artificial Intelligence
Lei Lu, Ping Wang, Yijie Cao
Summary: This paper proposes a novel part-level feature extraction method to enhance the discriminative ability of deep convolutional features for fine-grained vehicle recognition. By integrating the discriminative part features and modeling the coarse-to-fine relationship, the proposed method improves the representation of part-level features and achieves comparable performance with state-of-the-art algorithms.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Information Systems
Sheela Ramachandra, Suchithra Ramachandran
Summary: This paper proposes a Periocular recognition algorithm that utilizes region-specific and sub-image-based neighbor gradient feature extraction to achieve better recognition results. The proposed method segments the periocular region into sub-regions and extracts features using different algorithms. Experimental results demonstrate that the proposed method outperforms traditional algorithms on multiple datasets.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Hesam Omranpour, Vajihe Tirdad, Aref Misaghi
Summary: This paper presents an approach that identifies fingerprints by extracting geometric and statistical features of characteristic Minutiae. The proposed method divides the images into distinct regions by adding geometric features as preprocessing. Statistical parameters are applied to compute the general abstract of the features. Furthermore, this article introduces a similarity measure for identification tools used in pattern matching-based methods.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Sakshi Sharma, Sukhwinder Singh
Summary: This paper presents a small and efficient architecture for Indian sign language recognition, using transfer learning with MobileNetv2. Experimental results show that the proposed model outperforms the state-of-art method in terms of classification accuracy, computational complexity, and memory requirements.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Forestry
Yongke Sun, Guanben Du, Qizhao Lin, Lihui Zhong, Youjie Zhao, Jian Qiu, Yong Cao
Summary: This study proposed a secure labeling method that extracts the wood texture as a wood fingerprint to prevent label falsification. The method showed high recognition accuracy and robustness, making it suitable for tracing wood boards or logs to prevent illegal trading.
WOOD SCIENCE AND TECHNOLOGY
(2022)
Article
Engineering, Civil
Yujie Li, Jintong Cai, Quan Zhou, Huimin Lu
Summary: This paper introduces a new feature extraction network for point cloud segmentation tasks. By adding an encoder-decoder structure, the network can extract multi-scale local feature information and achieve better segmentation results in experiments.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Mathematics
Jiwei Wan, Huimin Zhao, Rui Li, Rongjun Chen, Tuanjie Wei
Summary: Given the interference of covariates in gait recognition, such as viewpoint and clothing, we propose an interval frame sampling method and an Omni-Domain Feature Extraction Network to weaken the influence of extrinsic variable changes and capture more joint dynamic information. The Omni-Domain Feature Extraction Network consists of three main modules: Temporal-Sensitive Feature Extractor, Dynamic Motion Capture, and Omni-Domain Feature Balance Module. Experimental results on two commonly used public gait datasets demonstrate the effectiveness and generalization ability of our method, achieving average rank-1 accuracies of 94.2% on CASIA-B and 90.5% on OU-MVLP.
Article
Computer Science, Artificial Intelligence
Yongke Sun, Qizhao Lin, Xin He, Youjie Zhao, Fei Dai, Jian Qiu, Yong Cao
Summary: In this paper, a novel deep-learning-based wood species recognition method was proposed. By using a 20X amplifying glass to acquire wood images, extracting features, optimizing, and identifying, the accuracy and generalization were significantly improved. Despite the limited amount of data, the method was successfully improved through transfer learning, demonstrating better performance in identifying 25 rare wood species.
INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Guanglong Du, Wenpei Zhou, Chunquan Li, Di Li, Peter X. Liu
Summary: This article proposes a hybrid neural network learning framework called CSFFN to detect a player's emotional states in real-time during a gaming process based on electroencephalogram (EEG) signals. CSFFN combines a convolutional neural network (CNN), a fuzzy neural network (FNN), and a recurrent neural network (RNN) to improve the accuracy and noise resistance in game emotion recognition. Experimental results show that CSFFN outperforms other methods in recognizing four emotional states (happiness, sadness, superiority, and anger).
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Article
Mathematical & Computational Biology
Shuai Cao, Biao Song
Summary: In this study, the flower category recognition problem is tackled using a deep learning model driven by visual attention mechanism, which enhances the learning and discriminating ability of the network through image augmentation and attention-driven approach. The proposed method achieves a recognition accuracy of 85.7% on a public dataset.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2021)
Article
Multidisciplinary Sciences
Sofien Gannouni, Arwa Aledaily, Kais Belwafi, Hatim Aboalsamh
Summary: Recognizing emotions using biological brain signals requires accurate signal processing and feature extraction methods. This study proposes a novel and adaptive channel selection method, along with the identification of epoch instants during emotions, to enhance the accuracy of the system. Experimental results show that the proposed method outperforms existing studies in multi-class emotion recognition with an average accuracy rate exceeding 89%. The method also shows improvement in accuracy rate when compared to existing algorithms dealing with 9 emotions, by 8%.
SCIENTIFIC REPORTS
(2021)
Article
Materials Science, Paper & Wood
Zihao Liu, Sulan Zhang, Xiaojun Jia, Jun Yang
Summary: This article proposes a novel feature extraction strategy based on improved blocked higher-order local auto-correlation (IBHLAC) for wood image texture. By dividing the entire wood image into grayscale-treated subdivisions, the memory consumption and complexity can be reduced. Additionally, by introducing morphology and affine transformation, the auto-correlation ability of the proposed method is enhanced.
Article
Computer Science, Artificial Intelligence
Yanping Chen, Weizhe Yang, Kai Wang, Yongbin Qin, Ruizhang Huang, Qinghua Zheng
Summary: The study proposed a neuralized feature engineering approach to enhance neural networks by manually designed features, which outperformed using neural networks or feature-based models alone, achieving better experimental results.
Article
Forestry
Zhengguang Wang, Zilong Zhuang, Ying Liu, Fenglong Ding, Min Tang
Summary: The study introduced machine vision technology and unsupervised learning technique to reduce labor costs and improve production efficiency through feature vector extraction and data dimension reduction for color classification. Texture recognition was achieved based on color classification, enhancing the quality stability of solid wood panels.
Article
Computer Science, Artificial Intelligence
Idil Isikli Esener
Summary: This paper analyzes the effects of various physiological signals on biometric identification and proposes a method for driver identification using wearable technology. The use of subspace-based features has been found to improve identification accuracy and reduce execution time. Furthermore, the foot galvanic skin response signal is shown to be more sensitive to identification and increased stress levels affect accuracy. These findings highlight the possibility of realizing driver-customized infotainment and vehicle security applications using a single wearable system.
Article
Engineering, Electrical & Electronic
Md Abdul Momin, Mohamad Haniff Junos, Anis Salwa Mohd Khairuddin, Mohamad Sofian Abu Talip
Summary: Efficient vehicle detection is crucial in intelligent transportation systems. This study aims to improve the conventional CNN model to achieve real-time detection on low-cost embedded hardware. A lightweight CNN model based on YOLOv4 Tiny is proposed, which adds an additional scale feature map to enhance detection accuracy. Experimental results show that the proposed model outperforms the conventional YOLOv4 Tiny and previous works in terms of mean average precision (mAP).
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Article
Mathematics
Sid Ahmed Ould Ahmed Mahmoud, El Moctar Ould Beiba, Fei Zuo
Summary: This work introduces and studies the concepts of n-quasi-m-hypercontractive and n-quasi-m-hyperexpansive operators on a Hilbert space. Some characterizations of these classes of operators are established, and many algebraic and spectral properties related to the members of these classes are extracted from these characterizations.
LINEAR & MULTILINEAR ALGEBRA
(2023)
Article
Clinical Neurology
Hani Chanbour, Lakshmi Suryateja Gangavarapu, Jeffrey W. Chen, Gabriel A. Bendfeldt, Iyan Younus, Mahmoud Ahmed, Steven G. Roth, Leo Y. Luo, Silky Chotai, Amir M. Abtahi, Byron F. Stephens, Scott L. Zuckerman
Summary: This study aims to investigate the incidence of unplanned readmission, identify associated risk factors, and determine the impact of unplanned readmission on long-term outcomes in patients undergoing surgery for cervical spine metastases. The results showed that 15% of patients experienced unplanned readmission, with 33% for surgical reasons and 67% for medical reasons. Preoperative radiation was associated with increased unplanned readmission rate, but readmission had no association with overall survival or local recurrence.
WORLD NEUROSURGERY
(2023)
Review
Computer Science, Information Systems
Ahmad A. M. Abushariah, Hua-Nong Ting, Mumtaz Begum Peer Mustafa, Anis Salwa Mohd Khairuddin, Mohammad A. M. Abushariah, Tien-Ping Tan
Summary: In this technological era, smart and intelligent systems with artificial intelligence techniques have impacted various aspects of daily life. Speech communication and interaction between humans and machines have become increasingly important, and numerous technologies, such as Automatic Speech Recognition (ASR), utilize speech as a means of interaction. However, research on ASR systems combining multiple languages is limited. This paper aims to provide a comprehensive background and fundamentals of bilingual ASR, discussing related works, research taxonomy, and open challenges. The study suggests that bilingual ASR with a deep learning approach is highly demanded and can open new research opportunities for language combinations.
Article
Medicine, Research & Experimental
Mahmoud S. Ahmed, Ayman B. Farag, Ian N. Boys, Ping Wang, Ivan Menendez-Montes, Ngoc Uyen Nhi Nguyen, Jennifer L. Eitson, Maikke B. Ohlson, Wenchun Fan, Matthew B. McDougal, Katrina Mar, Suwannee Thet, Francisco Ortiz, Soo Young Kim, Ashley Solmonson, Noelle S. Williams, Andrew Lemoff, Ralph J. DeBerardinis, John W. Schoggins, Hesham A. Sadek
Summary: The study aimed to identify therapeutic options for the COVID-19 pandemic through structure-based drug repurposing. FDA-approved drugs were screened for their inhibitory effects against SARS-CoV-2's main protease enzyme (Mpro) substrate-binding pocket. Five promising candidates, including atovaquone and mebendazole, demonstrated antiviral activity against SARS-CoV-2. Atovaquone was also found to potentially inhibit viral replication by targeting host purine metabolism.
BIOMEDICINE & PHARMACOTHERAPY
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Adam D. Yock, Mahmoud Ahmed, Sarah Masick, Manuel Morales-Paliza, Christien Kluwe, Ashwin Shinde, Austin Kirschner, Eric Shinohara
Summary: This study analyzed the clinical experience of CBCT-based daily online adaptive radiotherapy and demonstrated the impact of adaptive triggers on dosimetric and procedural outcomes. The results showed that daily online adaptive radiotherapy improved target coverage in multiple pelvic treatment sites, while the improvement in organ-at-risk metrics was randomly distributed.
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS
(2023)
Article
Mechanics
Ahmed Elhanafy, Yasser Abuouf, Samir Elsagheer, Shinichi Ookawara, Mahmoud Ahmed
Summary: Diagnostic technology based on magnetic fields is commonly used in medicine, but exposure to strong electromagnetic fields has adverse effects on patients. This study investigates the effects of external uniform magnetic fields on blood flow in healthy and diseased cases, and determines safe values for field strength. A three-dimensional non-Newtonian flow model is developed to investigate the effects of the magnetic field on shear rate and hematocrit. Numerical simulations are conducted at different field strengths and orientations, and results demonstrate the dominant effect of the magnetic field in the Y-direction.
Article
Engineering, Electrical & Electronic
Nur Athirah Zailan, Anis Salwa Mohd Khairuddin, Khairunnisa Hasikin, Mohamad Haniff Junos, Uswah Khairuddin
Summary: Garbage pollution is a growing global concern, and innovative solutions are necessary for its control. Obtaining visual information of floating garbage in rivers is crucial for the development of an efficient cleaner robot. Deep learning, which can learn high-level semantic features based on visual information, is essential for the detection and classification of different types of floating garbage. In this paper, an optimized You Only Look Once v4 Tiny model is proposed, which improves the spatial pyramid pooling, activation function, neural network, and hyperparameters to achieve better results. The proposed model shows a mean average precision of 74.89% with a size of 16.4 MB, making it the best trade-off among other models. It has promising results in terms of model size, detection time, and memory space, and is feasible to be embedded in low-cost devices.
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Article
Mathematics, Applied
Naeem Ahmad, E. S. Kamel, Sid Ahmed Ould Ahmed Mahmoud
Summary: In this work, we extend the concept of n-quasi-p-isometric operators from a single operator to multi-dimensional operators. We present new interesting results of these tuples of operators, which expand upon recently published works that focus on a single operator.
Article
Computer Science, Information Systems
Hai Chuan Liu, Joon Huang Chuah, Anis Salwa Mohd Khairuddin, Xian Min Zhao, Xiao Dan Wang
Summary: The intelligent campus surveillance system improves safety in school by recognizing abnormal behaviors. This study explores the challenges of video-based abnormal behavior recognition on campus and proposes a novel framework that models long-range temporal video structures and uses a global sparse uniform sampling strategy. The proposed method achieves competitive results in terms of recognition accuracy compared to other peer video recognition methods.
Article
Mining & Mineral Processing
Khaled Yassin, Mahmoud Ahmed, Mohamed Gamal Eldin Khalifa, Ayman Aly Hagrass
Summary: Dry magnetic separation is an effective technique for reducing iron in feldspar, especially in arid regions with limited water resources. This study conducted dry magnetic separation experiments on feldspar ore in the Wadi Umm Harjal area in Egypt, successfully removing iron content and improving the quality of the sample to meet industry specifications.
ARCHIVES OF MINING SCIENCES
(2023)
Article
Multidisciplinary Sciences
Sid Ahmed Ould Ahmed Mahmoud, Naeem Ahmad
Summary: In this paper, new classes of operators related to polynomially normal operators are introduced. The first class is m-quasi polynomially normal operators, which includes polynomially normal operators studied in recent papers. A bounded linear operator S is said to be an m-quasi polynomially normal operator if there exists a nontrivial polynomial P and a natural number m satisfying a certain equation. The second class is called polynomially C-normal operators, which includes C-normal operators studied in other papers. An operator S is called a polynomially C-normal operator if there exists a nontrivial polynomial P and a conjugation operator C satisfying a certain equation. The paper presents a detailed study of certain properties of the first class of operators and gives an introduction to the second class.
IRANIAN JOURNAL OF SCIENCE
(2023)
Article
Mathematics, Interdisciplinary Applications
Amin Sharafian, Jeevan Kanesan, Anis Salwa Mohd Khairuddin, Anand Ramanathan, Alireza Sharifi, Xiaoshan Bai
Summary: This paper presents a novel approach to designing a fixed-time fractional order observer for estimating the states of the dynamic model of HIV infection. The proposed approach combines output injection terminal sliding mode and RBF neural network strategies to achieve robust and efficient estimation. The results show accurate and efficient estimation of the states of the HIV model.
CHAOS SOLITONS & FRACTALS
(2023)
Article
Computer Science, Information Systems
Wen Yee Wong, Khairunnisa Hasikin, Anis Salwa Mohd Khairuddin, Sarah Abdul Razak, Hanee Farzana Hizaddin, Mohd Istajib Mokhtar, Muhammad Mokhzaini Azizan
Summary: A common difficulty in building prediction models with real-world environmental datasets is the skewed distribution of classes. This study evaluates the capability of machine learning algorithms in handling imbalanced water quality data and compares the performance of 10 algorithms. The results show that high-accuracy models are not always good in recall and sensitivity. The proposed stacked ensemble deep learning model performs well in the F1 score, achieving a balance between accuracy and completeness.
CMC-COMPUTERS MATERIALS & CONTINUA
(2023)
Article
Management
Mahmoud Ahmed
Summary: This paper investigates how women's representation on corporate boards affects financial reporting quality. The study finds that increasing women's representation on boards reduces undue CEO influence over the board and improves accounting quality. However, this effect is only significant in the long run and is attenuated for larger firms and larger boards.
INTERNATIONAL JOURNAL OF DISCLOSURE AND GOVERNANCE
(2023)
