Article
Automation & Control Systems
Guanjin Wang, Kup-Sze Choi, Jeremy Yuen-Chun Teoh, Jie Lu
Summary: This article introduces a new approach called DCOT-LS-SVMs, which is based on least-squares support vector machines and utilizes deep cross-output knowledge transfer. The approach improves the generalizability of LS-SVMs and simplifies the parameter tuning process. Experimental results on UCI datasets and a prostate cancer diagnosis case study demonstrate the effectiveness of the proposed approach.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Honorius Galmeanu, Razvan Andonie
Summary: This paper studies the problem of learning data sample distribution with concept drift. The author introduces a Weighted Incremental-Decremental SVM (WIDSVM) approach that can handle concept drift data streams with shifting windows, and it achieves significantly higher classification accuracy compared to traditional SVM methods.
Article
Computer Science, Artificial Intelligence
Pei-Yi Hao, Jung-Hsien Chiang, Yu-De Chen
Summary: This paper proposes a novel possibilistic classification algorithm using support vector machines (SVMs) to effectively handle uncertain information and improve classification performance. The algorithm aims at finding a maximal-margin fuzzy hyperplane based on possibility theory and solves a fuzzy mathematical optimization problem. The proposed algorithm retains the advantages of fuzzy set theory and SVM theory, and it is more robust for handling outliers. Experimental results demonstrate the satisfactory generalization accuracy and ability to describe inherent vagueness in the given dataset.
Article
Chemistry, Multidisciplinary
Honorius Galmeanu, Razvan Andonie
Summary: This study introduces a model called AIDSVM, which adjusts the width of the sliding window using the Hoeffding statistical test to adapt to concept drift. Experimental results show a significant improvement in accuracy when encountering concept drift, compared to similar drift detection models defined in the literature. AIDSVM is efficient, as it does not require retraining from scratch after the sliding window slides.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Artificial Intelligence
Imran Razzak, Mohamed Reda Bouadjenek, Raghib Abu Saris, Weiping Ding
Summary: Traditional support vector machines (SVMs) are sensitive to outliers and corrupted data, leading to a deterioration in classification performance. This article proposes an efficient Support Matrix Machine that performs matrix recovery and feature selection simultaneously. It can handle high-dimensional data with corrupted columns and recover an intrinsic matrix of higher rank under incoherence and ambiguity conditions. The objective function combines matrix recovery, low rank, and joint sparsity, and the method leverages structural information and intrinsic data structure. Experimental results show significant improvements in BCI, face recognition, and person identification datasets, especially in the presence of outliers.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Cybernetics
M. Tanveer, M. Tabish, Jatin Jangir
Summary: Analyzing unlabeled data and identifying underlying clustering principles is essential in various fields. Novel unsupervised machine learning algorithms, such as TBSVC, are developed for this purpose. However, TBSVC is sensitive to noise and lacks resampling stability. A new method, pinSTBSVC, uses a pinball loss function to improve sparsity and performance in plane-based clustering algorithms. Experimental results show that pinSTBSVC outperforms existing methods, demonstrating its effectiveness in real-world datasets and applications.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2022)
Article
Computer Science, Information Systems
M. Tanveer, S. Sharma, K. Muhammad
Summary: The proposed LS-LSTSVM addresses the shortcomings of TWSVM and LSTSVM by introducing a different Lagrangian function to eliminate the need for calculating inverse matrices, using the kernel trick directly for non-linear cases, and minimizing structural risk. These improvements aim to enhance classification accuracy on datasets, especially for large-scale problems.
ACM TRANSACTIONS ON INTERNET TECHNOLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Mehrnaz Ahmadi, Mehdi Khashei
Summary: Support vector machines (SVMs) are widely used in modeling, but traditional SVMs and fuzzy SVMs may not be sufficient for modeling both certain and uncertain patterns simultaneously. This paper proposes a generalized SVM that can effectively model both kinds of patterns to achieve more accurate wind speed forecasting results.
Article
Computer Science, Artificial Intelligence
Haiyan Chen, Ying Yu, Yizhen Jia, Bin Gu
Summary: This paper proposes an incremental learning algorithm ILTSVM based on the path following technique under the framework of infinitesimal annealing for training TSVM in handling large-scale data. Experimental results show that the proposed algorithm is the most effective and fastest method for training TSVM.
PATTERN RECOGNITION
(2023)
Article
Automation & Control Systems
Pedro Ribeiro Mendes Junior, Terrance E. Boult, Jacques Wainer, Anderson Rocha
Summary: When dealing with real-world recognition problems, it is often necessary to have classification methods that can handle unknown classes and reject samples not seen during training. Existing classifiers are mainly designed for closed-set scenarios, where all test samples are assumed to belong to known classes. However, in open-set scenarios, a test sample may not belong to any known class and must be properly rejected as unknown.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2022)
Article
Biochemical Research Methods
Ke Yan, Jie Wen, Jin-Xing Liu, Yong Xu, Bin Liu
Summary: The study proposed two novel algorithms, TSVM-fold and ESVM-fold, utilizing sequence similarity scores generated by multiple template-based methods for protein fold recognition prediction. Experimental results showed that these algorithms outperform some state-of-the-art methods in rigorous benchmark datasets.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2021)
Article
Computer Science, Information Systems
Eslam Eldeeb, Mohammad Shehab, Hirley Alves
Summary: The current random access allocation techniques for serving massive machine-type communication applications suffer from congestion and high signaling overhead. To address this issue, the third-generation partnership project introduced the use of fast uplink grant allocation to reduce latency and increase reliability. In this paper, a novel allocation method based on support vector machine is proposed, which prioritizes machine-type communication devices and uses traffic prediction and correction techniques to achieve efficient resource scheduling. The proposed method outperforms existing allocation schemes in terms of throughput and access delay when serving target massive and critical machine-type communication applications with limited resources.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Information Systems
Jayson P. Van Marter, Anand G. Dabak, Naofal Al-Dhahir, Murat Torlak
Summary: Ranging solutions for IoT localization applications aim to achieve high accuracy at low cost using Bluetooth low energy (BLE) technology. However, accurately measuring the distance with BLE poses challenges due to multipath components and model imperfections. To address this, we propose a data-driven SVR method that achieves decimeter-level accuracy with single antenna devices, outperforming the model-based MUSIC method which requires multiple antennas. Our method also proves robust in various multipath environments and offers computational complexity reduction compared to MUSIC.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Valentin Meiler, Jens Pfeiffer, Luca Bifano, Christoph Kandlbinder-Paret, Gerhard Fischerauer
Summary: We propose using electrical impedance spectroscopy (EIS) enhanced by machine learning (ML), particularly support vector machines (SVMs), as a noninvasive and in situ method for detecting microplastics in water. This method can provide a faster and cheaper alternative to laboratory measurements. We conducted stationary and dynamic measurements on water samples contaminated with different plastic concentrations and evaluated the results using SVMs. The classification accuracies for distinguishing plastic materials and particle sizes were over 98% and 91% respectively.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
M. Tanveer, A. Tiwari, R. Choudhary, M. A. Ganaie
Summary: This study proposes a novel large scale pinball twin support vector machine (LPTWSVM) to address the limitations of the twin support vector machines (TWSVMs), using a unique pinball loss function and improving model performance by eliminating matrix inversion calculation and minimizing structural risk.
Article
Geochemistry & Geophysics
Yi Huang, Jiangtao Peng, Yujie Ning, Qiang Cao, Weiwei Sun
Summary: This article proposes a new domain adaptation method called Distribution Alignment and Discriminative Feature Learning (DADFL) for hyperspectral image classification. By integrating category-discriminative information and structured prediction-based pseudolabeling, the DADFL method achieves better classification performance than existing methods in reducing distribution and subspace differences between domains.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Wenhui Hou, Na Chen, Jiangtao Peng, Weiwei Sun, Qian Du
Summary: This article proposes a semisupervised deep learning method for hyperspectral image classification, which integrates active learning, self-paced learning, and deep learning. It achieves better classification performance and reduces labeling cost.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geography, Physical
Yi Huang, Jiangtao Peng, Na Chen, Weiwei Sun, Qian Du, Kai Ren, Ke Huang
Summary: Wetlands are vital ecosystems, and using hyperspectral remote sensing technology for wetland mapping is crucial for preserving natural resources. However, challenges arise due to the high cost of labeled samples and differences in acquisition conditions. To address these difficulties, a spatial-spectral weighted adversarial domain adaptation network is proposed for cross-scene wetland mapping. The network aligns feature distributions, uses a weighted discriminator for source instance weighting, and incorporates a multi-classifier structure for improved classification performance. Experimental results demonstrate the superiority of the proposed method.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Yujie Ning, Jiangtao Peng, Quanyong Liu, Yi Huang, Weiwei Sun, Qian Du
Summary: This article explores the use of contrastive learning for cross-scene hyperspectral image classification. A category matching-based instance-to-instance contrastive learning framework is designed to learn discriminative feature embeddings by reducing feature distance. Experimental results demonstrate the effectiveness and feature discriminativeness of the proposed approach.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Kai Ren, Weiwei Sun, Xiangchao Meng, Gang Yang, Jiangtao Peng, Jingfeng Huang, Jiancheng Li
Summary: Image registration aims to eliminate geometric deviations between multisource data and promote their collaborative application. This article introduces a new robust method for hyperspectral (HS) and multispectral (MS) image registration based on common deep feature subspaces. The method effectively extracts common key points, weakens the differences in radiation and spatial texture information, and achieves high-precision matching. Experimental results demonstrate satisfactory registration performance and robustness compared to state-of-the-art methods.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Linzhou Yu, Jiangtao Peng, Na Chen, Weiwei Sun, Qian Du
Summary: This paper proposes a novel two-branch deeper graph convolutional network (TBDGCN) to extract both superpixel and pixel-level features for hyperspectral image classification. The TBDGCN model addresses the oversmoothing and overfitting problems by using the DropEdge technique and residual connection in the GCN branch, and captures attention-based spectral-spatial features through a mixed attention mechanism in the CNN branch. The features from both branches are fused for classification.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Quanyong Liu, Jiangtao Peng, Yujie Ning, Na Chen, Weiwei Sun, Qian Du, Yicong Zhou
Summary: In this article, a refined prototypical contrastive learning network for few-shot learning is proposed to address problems related to prototype instability and domain shift. By imposing triple constraints on prototypes of the support set, the prototypes are stabilized and refined. Additionally, a fusion training strategy is designed to alleviate the domain shift in few-shot learning.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Erxin Xie, Na Chen, Jiangtao Peng, Weiwei Sun, Qian Du, Xinge You
Summary: Recently, transformer-based networks have been used for hyperspectral image (HSI) classification. However, they lack the ability to effectively fuse different types of information in HSI. To address this, a novel semantic and spatial-spectral feature fusion transformer (S3FFT) network is proposed. The S3FFT method utilizes spatial attention, efficient channel attention (ECA), and a transformer-based module to extract and fuse advanced features for improved classification performance on HSI datasets.
CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY
(2023)
Article
Geochemistry & Geophysics
Weiwei Sun, Kai Ren, Xiangchao Meng, Gang Yang, Jiangtao Peng, Jiancheng Li
Summary: This study proposes a method for reconstructing remote sensing images with high temporal, spatial, and spectral resolution by fusing the temporal, spatial, and spectral information from multiple sources of remote sensing images. By utilizing tensor subspace decomposition and reconstruction networks, it effectively utilizes low-resolution hyperspectral images and high-resolution multispectral images. Experimental results demonstrate that the method achieves high-quality fusion results, exhibits comparable performance, and has robustness and practicality.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Wenhui Hou, Na Chen, Jiangtao Peng, Weiwei Sun
Summary: This letter proposes a new semi-supervised classification method called PALN, which integrates DL, AL, and PL into a framework. The method improves classification performance by selecting samples with high uncertainty and samples similar to prototypes.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Zhao Qiu, Jie Xu, Jiangtao Peng, Weiwei Sun
Summary: This article proposes a cross-channel dynamic spatial-spectral fusion Transformer (CDSFT) method for addressing the problem of ignoring spatial information variations in hyperspectral image classification. By combining CNN and Transformer, the CDSFT achieves more effective feature extraction and global modeling, thereby improving the classification performance.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Quanyong Liu, Jiangtao Peng, Na Chen, Weiwei Sun, Yujie Ning, Qian Du
Summary: This article proposes a cross-domain few-shot learning method for hyperspectral image classification, which improves classification performance through contrastive learning and prototype self-refinement strategy, and performs domain adaptation with a local discriminative domain adaptation method. Experimental results demonstrate that this method outperforms existing methods in hyperspectral image classification.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Weiwei Liu, Kai Liu, Weiwei Sun, Gang Yang, Kai Ren, Xiangchao Meng, Jiangtao Peng
Summary: This article proposes an HSI classification method based on self-supervised learning of spectral masking (SSLSM). The method consists of self-supervised pretraining and fine-tuning steps, which improve classification accuracy and has been verified on multiple datasets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Yujie Ning, Jiangtao Peng, Lin Sun, Yi Huang, Weiwei Sun, Qian Du
Summary: In this article, an adaptive local discriminant analysis and distribution matching (ALDADM) method is proposed for domain adaptation in hyperspectral image (HSI) classification. ALDADM extracts discriminative and robust features using adaptive local discriminant analysis, and reduces domain shift through distribution matching and subspace alignment.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Kai Ren, Weiwei Sun, Xiangchao Meng, Gang Yang, Jiangtao Peng, Jingfeng Huang
Summary: This article presents a locally optimized image segmentation fusion (LOISF) framework for HS super-resolution reconstruction, which preserves spatial details and texture while improving image quality in LR HS and HR MS images through the construction of a joint fusion model and a convex optimization solution.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)