Article
Computer Science, Interdisciplinary Applications
Varanasi Satya Sreekanth, Karnam Raghunath, Deepak Mishra
Summary: Atmospheric Gravity Waves play a significant role in Middle Atmosphere Dynamics, and the breaking of Gravity Waves leads to turbulence. However, the accuracy and sparsity of Wind Velocity measuring instruments at the altitude of interest pose a problem for confirming the breaking of Atmospheric Gravity Waves. In this study, we propose a solution using Dictionary Learning and Deep Learning methods to detect Wave Breaking events from atmospheric temperature perturbations, and the effectiveness of this method is demonstrated through a case study using satellite data and validated with ground-based instruments.
COMPUTERS & GEOSCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Hong Peng, Cancheng Li, Jinlong Chao, Tao Wang, Chengjian Zhao, Xiaoning Huo, Bin Hu
Summary: This study proposes a novel sparse representation-based epileptic seizure classification method based on dictionary learning, which is evaluated on public EEG databases. The new method shows higher automation and recognition rates compared to traditional methods.
Article
Computer Science, Artificial Intelligence
Markus Viljanen, Antti Airola, Tapio Pahikkala
Summary: Pairwise learning is a supervised learning setting for predicting outcomes of object pairs. This review focuses on pairwise kernels that incorporate prior knowledge about object relationships. The use of a generalized vec trick algorithm allows for faster computation of the kernels, enabling their application to larger datasets.
Article
Engineering, Biomedical
Jijun Tong, Kai Li, Wenting Lin, Shudong Xia, Ali Anwar, Lurong Jiang
Summary: The study introduces a fully automatic method for detecting lumen contours in coronary artery IVUS images, utilizing texture feature vectors, dictionary construction, and classification algorithms. Evaluation using public indicators demonstrates that the proposed method outperforms others in detection accuracy, showing good performance overall.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Engineering, Electrical & Electronic
John Joseph Hall, Christopher Robbiano, Mahmood R. Azimi-Sadjadi
Summary: This paper presents a new method for incrementally updating features in adaptive classification for pattern discrimination in varying conditions. The method leverages a fast eigendecomposition algorithm for symmetric Arrowhead matrices in the Nystrom-approximated linearized kernel embedding (LKE) for improved performance. It can also be applied to kernel principal component analysis (KPCA) or similar problems with good results.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Inzamam Mashood Nasir, Mudassar Raza, Siti Maghfirotul Ulyah, Jamal Hussain Shah, Norma Latif Fitriyani, Muhammad Syafrudin
Summary: This study proposes a network-level fusion method that extracts unique features by combining multiple pre-trained models effectively. Five fusion strategies, including sum, max, concatenation, convolutional, and bilinear fusion, are used to fuse three pre-trained models. Finally, an optimization method is used to extract descriptors, and the proposed model is evaluated on four publicly available datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Di Wu, PinYi Zhao, Qin Wan
Summary: In this paper, a novel classification framework called WELM-ERDDL is proposed to address the issues of unsatisfied performance and high time consumption in image classification. The proposed method combines WELM and ERDDL to achieve better efficiency compared to other state-of-the-art algorithms.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Youkabed Amiri, Hesam Omranpour
Summary: This paper presents a spatial learning concept for classifying EEG motor imagery signals. The proposed method increases and then reduces the data dimensions to learn an efficient space for signal classification. Experimental results show that the method achieves high classification accuracy and robustness on the BCI Competition dataset.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Mathematics, Applied
Eun-Ji Lee, Jae-Hwan Jhong
Summary: This study proposed a function estimation method with change point detection using truncated power spline basis and L1-norm penalty. Two computational algorithms were introduced and compared for solving the proposed estimators. Numerical studies were conducted using simulation and real data analysis to validate the performance of the method.
Article
Computer Science, Information Systems
Fatemeh Zamani, Mansour Jamzad, Hamid R. Rabiee
Summary: The paper proposed an MKL-SRC method with non-fixed kernel weights, which can compute an atom-specific multiple kernel dictionary in the training phase for classifying test images. The effectiveness of the proposed approach was demonstrated through experimental results.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2021)
Article
Multidisciplinary Sciences
Weina Wang, Shuangyong Li, Jiapeng Shao, Huxidan Jumahong
Summary: Deep learning-based object detection methods have shown great improvement in performance, but the use of small kernel convolution makes it difficult to obtain semantic features, leading to problems such as wrong detection, missing detection, and repeated detection. To address these issues, we propose LKC-Net, a large kernel convolution object detection network that enhances feature capture and utilizes a vast receptive field attention mechanism.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Artificial Intelligence
Zhou Huang, Huai-Xin Chen, Tao Zhou, Yun-Zhi Yang, Chang-Yin Wang, Bi-Yuan Liu
Summary: A novel saliency detection model based on Contrast weighted Dictionary Learning (CDL) is proposed in this paper for VHR optical RS images, which effectively measures saliency by combining the coefficients of the sparse representation and reconstruction errors, and integrates multiple saliency maps using a fusion method based on global gradient optimization.
PATTERN RECOGNITION
(2021)
Article
Psychology, Clinical
Neusa Aita Agne, Caroline Gewehr Tisott, Pedro Ballester, Ives Cavalcante Passos, Ygor Arzeno Ferrao
Summary: This study used a machine learning algorithm to identify predictors of suicide attempts in patients with OCD. The relevant predictors found were previous suicide planning, previous suicide thoughts, lifetime depressive episode, and intermittent explosive disorder. This is the first study to evaluate suicide risk factors in OCD patients using machine learning algorithms.
PSYCHOLOGICAL MEDICINE
(2022)
Article
Geosciences, Multidisciplinary
Huajin Li, Qiang Xu, Yusen He, Xuanmei Fan, He Yang, Songlin Li
Summary: This study proposed a deep learning framework based on long short-term memory to model and predict sharp deformation of landslides, demonstrating its accuracy in identifying future sharp deformations. The use of Hurst exponent in prediction errors helped reveal abnormal patterns in sharp deformation, providing valuable support for on-site risk analysis and decision-making by geological engineers.
GEOMATICS NATURAL HAZARDS & RISK
(2021)
Article
Computer Science, Information Systems
Linrui Shi, Zheng Zhang, Zizhu Fan, Chao Xi, Zhengming Li, Gaochang Wu
Summary: Dictionary learning is an efficient method to learn important features of data. The proposed Kernel Fisher Dictionary Transfer Learning (KFDTL) algorithm addresses the issue of nonlinear information extraction from large-scale and high-dimensional datasets. By mapping samples to high-dimensional space and using dictionary learning algorithm, essential features are learned, and feature transfer learning is performed for label prediction of target samples. Experimental results on public image datasets demonstrate the effectiveness of the proposed method.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)