Article
Computer Science, Artificial Intelligence
Muhammad Waqas, Muhammad Atif Tahir, Salman A. Khan
Summary: Multi-instance learning (MIL) allows predictive algorithms to use complex data representation. This paper proposes a fuzzy subspace clustering approach and an ensemble-based variant of Fisher vector (FV) encoding, named FCBE-miFV, to tackle the challenges in MIL, such as hypothesis space complication and robust instance selection. The proposed algorithm improves model performance by incorporating essential instances in the bag encoding process.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Guanli Yue, Yanpeng Qu, Longzhi Yang, Changjing Shang, Ansheng Deng, Fei Chao, Qiang Shen
Summary: Fuzzy clustering is a method that uses partial memberships to decompose data into clusters, and it demonstrates comparable performance in knowledge exploitation when dealing with incomplete information. This article proposes a new fuzzy-rough intrigued harmonic discrepancy clustering (HDC) algorithm that effectively handles complex data distribution and improves clustering performance.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Fang Wan, Qixiang Ye, Tianning Yuan, Songcen Xu, Jianzhuang Liu, Xiangyang Ji, Qingming Huang
Summary: This paper proposes a Multiple Instance Differentiation Learning (MIDL) method for instance-level active learning, which unifies instance uncertainty with image uncertainty for informative image selection. Extensive experiments on commonly used object detection datasets validate that MIDL outperforms other state-of-the-art methods, especially when the labeled sets are small.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Guanli Yue, Ansheng Deng, Yanpeng Qu, Hui Cui, Jiahui Liu
Summary: Ensemble clustering involves constructing multiple base clusterings to achieve fast clustering under abundant computing resources. By integrating multiple clustering algorithms, ensemble clustering has stronger robustness and applicability compared to a single clustering algorithm. However, most ensemble clustering algorithms treat each base clustering result equally and overlook the differences between clusters. This paper proposes a novel fuzzy-rough induced spectral ensemble approach to enhance clustering performance. The proposed approach differentiates the significance of clusters and induces the unacceptable degree and reliability of clusters formed in base clustering based on fuzzy-rough lower approximation. Experimental results demonstrate the effectiveness of the proposed approach, which outperforms state-of-the-art ensemble clustering algorithms and base clustering, highlighting the superiority of this novel algorithm.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2023)
Review
Computer Science, Artificial Intelligence
Wanting Ji, Yan Pang, Xiaoyun Jia, Zhongwei Wang, Feng Hou, Baoyan Song, Mingzhe Liu, Ruili Wang
Summary: Feature selection is a key method for data preprocessing in data mining tasks, aiming to select a feature subset based on evaluation criteria. Fuzzy rough set theory has been proven to be ideal for dealing with uncertain information in feature selection. This article provides a comprehensive review of fuzzy rough set theory and its applications, discussing challenges in feature selection methods.
WILEY INTERDISCIPLINARY REVIEWS-DATA MINING AND KNOWLEDGE DISCOVERY
(2021)
Article
Computer Science, Artificial Intelligence
Wenqian He, Shihu Liu, Weihua Xu, Fusheng Yu, Wentao Li, Fang Li
Summary: This paper introduces a novel method of graph data clustering based on rough set theory. The method partitions the graph data into overlapping subgraph data by constructing an optimization model and updating fuzzy membership degree and cluster center. Experimental results demonstrate that the proposed method outperforms existing clustering approaches to some extent.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Computer Science, Information Systems
Yan Sun, Fu-Gui Shi
Summary: There has been a growing interest in the development of lattice-valued rough set theory in recent years. This framework includes various types of rough sets, such as intuitionistic, interval-valued, neutrosophic, and Pythagorean fuzzy rough sets. Additionally, representations of fuzzy rough approximation operators based on a completely distributive lattice are provided, extending the existing results for the case when the lattice is [0,1]. It is shown that the representations of intuitionistic and interval-valued fuzzy rough approximation operators proposed by Zhou and Sun are special cases of the proposed representations.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Jin Ye, Jianming Zhan, Weiping Ding, Hamido Fujita
Summary: This paper redefines fuzzy beta-neighborhood operators to achieve reflexivity, constructs an ITFRS model based on these reflexive operators, and proposes a new decision-making method for multi-criteria decision-making problems.
INFORMATION SCIENCES
(2021)
Article
Chemistry, Multidisciplinary
Chih-Ming Chen, Sheng-Chieh Chang, Chen-Chia Chuang, Jin-Tsong Jeng
Summary: This study introduces a new RIPFCM clustering algorithm that can better handle symbolic interval data in noisy environments and with data overlapping problems. The algorithm is applied on smartphones, expanding the computing power of smartphones and showing potential for new applications in SDA.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Theory & Methods
Junli Zhou, Fasheng Xu, Yanyong Guan, Hongkai Wang
Summary: This paper introduces three new types of fuzzy covering-based rough set models, defined based on membership degree and membership function in different approximation spaces. The study includes properties of the models and their relationship with traditional rough sets.
FUZZY SETS AND SYSTEMS
(2021)
Article
Automation & Control Systems
Akin Osman Atagun, Huseyin Kamaci
Summary: This paper introduces a new uncertainty modelling concept called strait fuzzy set, which brings new perspectives to fuzzy mathematics. Strait fuzzy sets allow objects/points to be graded with fuzzy membership intervals instead of exact values. The paper also studies the basic operations and properties of strait fuzzy sets, introduces the concept of strait fuzzy rough set, and proposes similarity approaches for measuring similarity rates of vaccines against influenza viruses.
INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE
(2023)
Article
Multidisciplinary Sciences
A. A. Abdallah, O. R. Sayed, E. El-Sanousy, Y. H. Ragheb Sayed, M. N. Abu-Shugair, Salahuddin
Summary: In this article, two new rough set models, optimistic and pessimistic multi-granulation double fuzzy rough sets, based on the concept of double fuzzy relations, are introduced and discussed. The study focuses on the lower and upper approximations, which generalize the conventional rough set model. The development of the multi-granulation double fuzzy rough set model is suggested to be significant for the generalization of the rough set model.
Article
Computer Science, Information Systems
Chenxia Jin, Jusheng Mi, Fachao Li, Meishe Liang
Summary: This study explores and applies the fusion of probabilistic hesitant fuzzy sets (PHFSs) and rough sets in uncertain multi-criteria decision-making (MCDM). It introduces an advanced method to obtain normalized PHFSs (NPHFSs), establishes a novel probabilistic hesitant fuzzy rough set (PHFRS) model, and proposes a fuzziness-based objective weight determination method. The effectiveness of the proposed method is demonstrated through experimental comparisons.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Software Engineering
Leiyu Tang, Chenxi Wang, Shudong Wang, Guoliang Dong, Jiancong Fan
Summary: Rough set theory is an important method for analyzing data with uncertainties, able to acquire knowledge by the indistinguishable relationship among data objects without any prior knowledge. The novel fuzzy clustering algorithm based on rough set and inhibitive factor shows potential application value in data mining by improving the convergence speed while guaranteeing the clustering effect.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2021)
Article
Computer Science, Information Systems
Changzhong Wang, Yang Huang, Weiping Ding, Zehong Cao
Summary: Fuzzy rough sets combined with the concept of self-information are used to construct four uncertainty measures to evaluate the classification ability of attribute subsets. The fourth measure, relative decision self-information, is proven to be better for attribute reduction. A greedy algorithm is designed for attribute reduction, and the effectiveness of the method is validated through experimental results.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Wenhui Wu, Yujie Chen, Ran Wang, Le Ou-Yang
Summary: This paper proposes a semi-supervised self-representative kernel concept factorization (S3RKCF) method that integrates adaptive kernel learning and low-dimensional data representation learning into a unified model. An adaptive local geometric structure is acquired in the KCF-induced self-representation space to facilitate data representation learning. Limited supervisory information is imposed as constraints to enhance the discriminability of data representation. The proposed S3RKCF outperforms state-of-the-art methods in clustering and classification tasks according to experimental results.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Mingliang Zhou, Xuekai Wei, Weijia Jia, Sam Kwong
Summary: In this paper, a joint decision tree and visual feature optimization rate control scheme for ultrahigh-definition (UHD) versatile video coding (VVC) is proposed. The scheme includes a new rate-distortion (R-D) model for UHD videos, a decision-tree-based multiclass classification scheme, and a convex optimization algorithm. Experimental results show that compared to other state-of-the-art algorithms, the proposed method achieves significant bit rate reductions while maintaining a given peak signal-to-noise ratio (PSNR) or structural similarity index measure (SSIM).
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Shuyue Chen, Ran Wang, Jian Lu
Summary: Multi-label Active Learning (MLAL) is an effective method that improves the performance of multi-label classifiers with less annotation effort. This paper proposes a deep reinforcement learning (DRL) model to explore a general evaluation method for MLAL and addresses label correlation and data imbalanced problems using a self-attention mechanism and a reward function. Experimental results show that the DRL-based MLAL method achieves comparable results to other methods reported in the literature.
Article
Computer Science, Artificial Intelligence
Jian-Yu Li, Zhi-Hui Zhan, Jin Xu, Sam Kwong, Jun Zhang
Summary: This article proposes a novel estimation of distribution algorithm (EDA), named surrogate-assisted hybrid-model EDA (SHEDA), for efficient hyperparameters optimization. The algorithm design includes hybrid-model EDA, orthogonal initialization strategy, and surrogate-assisted multi-level evaluation method. Experimental results show that SHEDA is very effective and efficient for hyperparameters optimization on widely used classification benchmark problems.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Si-Chen Liu, Zong-Gan Chen, Zhi-Hui Zhan, Sang-Woon Jeon, Sam Kwong, Jun Zhang
Summary: This article addresses the job-shop scheduling problem with multiple objectives, including completion time, total tardiness, advance time, production cost, and machine loss. A multiple populations for multiple objectives genetic algorithm (MPMOGA) is proposed to optimize these objectives simultaneously. The MPMOGA algorithm utilizes an archive sharing technique and an archive update strategy to improve the quality and diversity of the solutions. Experimental results show that MPMOGA outperforms other state-of-the-art algorithms on most test instances.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Automation & Control Systems
Runmin Cong, Ning Yang, Chongyi Li, Huazhu Fu, Yao Zhao, Qingming Huang, Sam Kwong
Summary: This article proposes a global-and-local collaborative learning architecture (GLNet) to effectively extract interimage correspondence in co-salient object detection. The GLNet utilizes global and local correspondence modeling, pairwise correlation transformation, and correspondence aggregation to enhance the comprehensive interimage collaboration cues. The evaluation results demonstrate the superiority of GLNet over state-of-the-art competitors.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Engineering, Electrical & Electronic
Yue Qian, Junhui Hou, Qijian Zhang, Yiming Zeng, Sam Kwong, Ying He
Summary: This paper presents MOPS-Net, a deep learning-based method for compact representation of 3D point clouds using matrix optimization. It achieves favorable performance in various tasks and exhibits robustness to noisy data.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Yi Chen, Meng Wang, Shiqi Wang, Zhangkai Ni, Sam Kwong
Summary: In this paper, a rate control scheme is proposed for screen content video coding in the VVC standard. The method relies on pre-analysis to obtain content information and incorporates complexity-aware rate models and distortion models to achieve optimal bit allocations. Experimental results demonstrate the effectiveness of the proposed method in improving coding performance.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Jielian Lin, Aiping Huang, Tiesong Zhao, Xu Wang, Sam Kwong
Summary: In this paper, a solution is proposed to address the bit allocation problem in VVC video compression by formulating it as a Nash equilibrium problem. By introducing ?-domain RD models, a constrained optimization problem is derived and solved using a Newton method and Nash equilibrium. Experimental results demonstrate the effectiveness and superiority of the proposed method.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Yu Zhou, Yan Qiu, Sam Kwong
Summary: In contrast to traditional feature selection methods, local feature selection methods partition the sample space and obtain feature subsets for each local region. However, most existing local feature selection algorithms lack a problem-specific objective function and instead use a distance-like objective function, leading to limited classification performance. In this article, we propose a novel objective function called region purity (RP) for local feature selection. To solve this problem, we use an improved nondominated sorting genetic algorithm III and develop a regional feature sharing strategy. Experimental results on various datasets demonstrate the effectiveness of our proposed RP-LFS. Compared to other state-of-the-art feature selection and local feature selection algorithms, RP-LFS achieves competitive classification accuracy while reducing the feature subset size.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
Ran Wang, Shuyue Chen, Yu Yu
Summary: Version space is a crucial concept in supervised learning, but its application in multi-label active learning has not been explored. This paper extends the version space theory from single-label scenario to multi-label scenario, establishes a spatial structure for the multi-label version space, and proposes a simplified representation and a new multi-label active learning algorithm. The algorithm is further enhanced by addressing the issue of class imbalance in multi-label data. Experimental comparisons demonstrate the feasibility and effectiveness of the proposed methods.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Runmin Cong, Wenyu Yang, Wei Zhang, Chongyi Li, Chun-Le Guo, Qingming Huang, Sam Kwong
Summary: Due to light absorption and scattering in water, underwater images often suffer from degradation issues such as low contrast, color distortion, and blurriness, making underwater understanding tasks more challenging. To address this, the study proposes a physical model-guided GAN model called PUGAN, which combines the advantages of GANs in visual aesthetics and physical model-based methods in scene adaptability. The proposed model includes a Parameters Estimation subnetwork for physical model inversion and a Two-Stream Interaction Enhancement subnetwork with a Degradation Quantization module. Dual-Discriminators are also designed for adversarial constraint to improve authenticity and visual aesthetics.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Xingran Liao, Xuekai Wei, Mingliang Zhou, Sam Kwong
Summary: This paper proposes a deep order statistical similarity (DOSS) FR-IQA model to evaluate content-misaligned image pairs encountered in image reconstruction and texture synthesis tasks. DOSS compares the order statistics of deep features in the reference and distorted images to output perceptual quality scores. It mimics the human visual system's behavior and possesses advanced texture perception capability, producing superior quality assessment results on various texture synthesis algorithms.
IEEE TRANSACTIONS ON BROADCASTING
(2023)
Article
Engineering, Electrical & Electronic
Haifeng Guo, Sam Kwong, Chuanmin Jia, Shiqi Wang
Summary: Most deep learning-based video compression frameworks rely on motion estimation and compensation, but the artifacts of warped frames limit the performance. In this work, we propose enhanced motion compensation to reduce error propagation. We incorporate a designed convolutional neural network into Open DVC as the enhancement network, and optimize the framework with a single loss function considering the trade-off between bit cost and frame quality. Experimental results show that our model achieves significant bit savings and outperforms Open DVC in terms of PSNR and bit rate savings.
IEEE SIGNAL PROCESSING LETTERS
(2023)
Article
Computer Science, Hardware & Architecture
Guang Han, Chen Cao, Jixin Liu, Sam Kwong
Summary: This article proposes a new Anchor-free Tracker based on Space-time Memory Network (ATSMN) to solve the appearance problems in object tracking. By utilizing space-time memory network, memory feature fusion network, and transformer feature cross fusion network, the tracker can effectively use temporal context information and better adapt to appearance changes, achieving accurate classification and regression results. Extensive experimental results show that ATSMN outperforms other advanced trackers on challenging benchmarks.