检索结果-内蒙古大学图书馆

FSPDF: Few-shot learning with progressive dual-domain feature fusion via self-supervised learning

KNOWLEDGE-BASED SYSTEMS 2025年 317卷

作者： Li, Dongqing Jin, Jie Zou, Linhua Zhao, Hong Minnan Normal Univ Sch Comp Sci Zhangzhou 363000 Fujian Peoples R China Minnan Normal Univ Key Lab Data Sci & Intelligence Applicat Zhangzhou 363000 Fujian Peoples R China

Few-shot learning (FSL) aims to develop models that can generalise to new tasks using only a few labelled examples. Recently, feature fusion methods have achieved great success in FSL by aggregating features from different sources. However, these methods generally rely on efficient feature extractors to capture the intrinsic patterns of the data and focus solely on single-domain spatial features. In this paper, we propose a progressive dual-domain feature fusion strategy via self-supervised learning (FSPDF). FSPDF consists of dual-domain feature learning (DFL) and dual-domain feature fusion (DFF) and leverages progressive dual-domain fusion to address the issue of insufficient feature representation. The DFL strategy obtains an efficient feature extractor through a dual-domain self-supervised module, optimising generalisation to new data and fully extracting dual-domain features. The DFF strategy leverages the previously obtained feature extractor and incorporates discrete wavelet transform to enrich and fuse dual-domain features, providing a more diverse feature representation. The effectiveness of FSPDF is demonstrated across three commonly used benchmark FSL datasets. For instance, in the 5-way 1-shot setting, FSPDF improves performance by 3.47% on CUB-200-2011 and 1.05% on miniImageNet. Code is available at https://***/fhqxa/FSPDF.

关键词： Few-shot learning Feature fusion Discrete wavelet transform Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Mg-SubAgg: Multi-granularity Subgraph Aggregation with topology for GNN

引用

INFORMATION SCIENCES 2024年 677卷

作者： Zhang, Xiaoxia Ye, Mengsheng Zhang, Yun Liu, Qun Wang, Guoyin Wu, Kesheng Chongqing Univ Posts & Telecommun Chongqing Key Lab Computat Intelligence Chongqing 400065 Peoples R China Chongqing Univ Posts & Telecommun Key Lab Big Data Intelligent Comp Chongqing 400065 Peoples R China Chongqing Univ Posts & Telecommun Key Lab Cyberspace Big Data Intelligent Secur Minist Educ Chongqing 400065 Peoples R China Lawrence Berkeley Natl Lab Sci Data Div Berkeley CA 94720 USA

Although graph neural networks (GNNs) work well on graph data, they are black-box models that lack of reliable explanations for their predictions. We propose a multi-granularity subgraph aggregation method based on graph topology to explain GNNs. Specifically, given a trained GNN model and an input graph, our method constructs a subgraph by heuristics from fine-grained to coarse-grained and sorts the subgraph nodes to obtain subgraph and node-level explain. Furthermore, we propose an improved Shapley value as a heuristic function for the search algorithm, which strikes a balance between the time complexity and accuracy. Finally, experimental results on both synthetic and real datasets demonstrate that our method achieves best performance on seven datasets, quantifying the influence of individual nodes on prediction results and providing more reliable explanations.

关键词： Graph Neural Network (GNN) Explainability of GNN Subgraph Aggregation Improved Shapley value Node importance

来源：评论

学校读者我要写书评

暂无评论

Matrix-based incremental local feature selection with dynamic covering granularity

引用

APPLIED intelligence 2025年第5期55卷 1-18页

作者： Shi, Qi Zhang, Yan-Lan Minnan Normal Univ Sch Comp Sci Zhangzhou 363000 Fujian Peoples R China Fujian Prov Univ Key Lab Data Sci & Intelligence Applicat Zhangzhou 363000 Fujian Peoples R China

Multigranulation rough set is composed of a set of granularities, providing a theoretical framework for solving problems from a multigranulation perspective. Feature selection aims to find the minimal set of attributes that does not compromise the overall classification capability. It has significant applications in the field of information processing. However, in practical application environments, the granularities in information systems often evolve dynamically over time. To address this scenario, an incremental feature selection algorithm for data with changing granularities in local multigranulation neighborhood covering rough sets is proposed. Firstly, the method of local related family is introduced, relationships between matrix operations of local related sets and those of approximate sets are discussed, and feature selection is studied using matrix methods. Subsequently, two matrix-based incremental feature selection algorithms are proposed for the cases where granularity structures in the data are added or deleted due to feature changes. Experiments on six datasets from UCI are then conducted to evaluate the performance of the proposed algorithms. The experimental results demonstrate that the two proposed incremental feature selection algorithms are highly effective.

关键词： Neighborhood covering granularity Feature selection Incremental algorithm Dynamic information system Local multigranulation rough sets Matrix

来源：评论

学校读者我要写书评

暂无评论

Matrix-based local multigranulation reduction for covering decision information systems

引用

INTERNATIONAL JOURNAL OF APPROXIMATE REASONING 2025年 181卷

作者： Jiang, Tao Zhang, Yan-Lan Minnan Normal Univ Sch Comp Sci Zhangzhou 363000 Fujian Peoples R China Fujian Prov Univ Key Lab Data Sci & Intelligence Applicat Zhangzhou 363000 Fujian Peoples R China

Attribute reduction has become an essential step in pattern recognition and machine learning tasks. As an extension of the classical rough set, the covering rough set has garnered considerable attention in both theory and application. A matrix-based method for computing local covering optimistic approximation sets and local optimistic multigranulation reductions based on covering rough set in covering decision information systems (CDISs) is proposed in this paper. Firstly, we introduce a matrix representation along with its associated operations to compute the local covering optimistic approximation sets and the local positive regions of the CDISs. Subsequently, local optimistic discernibility matrices and local optimistic discernibility functions are constructed for the CDISs. By performing disjunction and conjunction operations on these local optimistic discernibility matrices, all local optimistic multigranulation reductions of the CDISs can be accurately obtained. In addition, an algorithm is developed using the local optimistic discernibility matrix to compute a suboptimal minimal local optimistic multigranulation reduction. Finally, to verify the effectiveness and feasibility of the proposed method, numerical experiments are conducted on 6 UCI datasets.

关键词： Local optimistic multigranulation reduction Matrix Covering decision information system

来源：评论

学校读者我要写书评

暂无评论

Efficient gradient descent algorithm with anderson acceleration for separable nonlinear models

引用

NONLINEAR DYNAMICS 2025年第10期113卷 11371-11387页

作者： Chen, Guang-Yong Lin, Xin Xue, Peng Gan, Min Fuzhou Univ Coll Comp & Data Sci Fuzhou 350116 Peoples R China Minist Educ Fujian Key Lab Network Comp & Intelligent Informat Key Lab Intelligent Metro Univ Fujian Fuzhou 350108 Peoples R China Minist Educ Engn Res Ctr Big Data Intelligence Fuzhou 350108 Peoples R China Qingdao Univ Coll Comp Sci & Technol Qingdao 266071 Peoples R China

Separable nonlinear models are pervasively employed in diverse disciplines, such as system identification, signal analysis, electrical engineering, and machine learning. Identifying these models inherently poses a non-convex optimization challenge. While gradient descent (GD) is a commonly adopted method, it is often plagued by suboptimal convergence rates and is highly dependent on the appropriate choice of step size. To mitigate these issues, we introduce an augmented GD algorithm enhanced with Anderson acceleration (AA), and propose a Hierarchical GD with Anderson acceleration (H-AAGD) method for efficient identification of separable nonlinear models. This novel approach transcends the conventional step size constraints of GD algorithms and considers the coupling relationships between different parameters during the optimization process, thereby enhancing the efficiency of the solution-finding process. Unlike the Newton method, our algorithm obviates the need for computing the inverse of the Hessian matrix, simplifying the computational process. Additionally, we theoretically analyze the convergence and complexity of the algorithm and validate its effectiveness through a series of numerical experiments.

关键词： Anderson acceleration Separable nonlinear problem Robust parameter estimation Hierarchical identification algorithm

来源：评论

学校读者我要写书评

暂无评论

CFDRN: A cognition-Inspired Feature Decomposition and Recombination Network for Dysarthric Speech Recognition

引用

IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 2023年 31卷 3824-3836页

作者： Lin, Yuqin Wang, Longbiao Yang, Yanbing Dang, Jianwu Tianjin Univ Coll Intelligence & Comp Tianjin Key Lab Cognit Comp & Applicat Tianjin 300350 Peoples R China Huiyan Technol Tianjin Co Ltd Tianjin 300350 Peoples R China

As an essential technology in human-computer interactions, automatic speech recognition (ASR) ensures a convenient life for healthy people;however, people with speech disorders, who truly need support from such a technology, have experienced difficulties in the use of ASR. Disordered ASR is challenging because of the large variabilities in disordered speech. Humans tend to separately process different spectro-temporal features of speech in the left and right hemispheres of their brain, showing significantly better ability in speech perception than machines, especially in disordered speech perception. Inspired by human speech processing, this article proposes a cognition-inspired feature decomposition and recombination network (CFDRN) for dysarthric ASR. In the CFDRN, slow- and rapid-varying temporal processors are designed to decompose features into stable and changeable features, respectively. A gated fusion module was developed to selectively recombine the decomposed features. Moreover, this study utilised an adaptation approach based on unsupervised pre-training techniques to alleviate data scarcity issues in dysarthric ASR. The CFDRNs were added to the layers of the pre-trained model, and the entire model is adapted from normal speech to disordered speech. The effectiveness of the proposed method was validated on the widely used TORGO and UASpeech dysarthria datasets under three popular unsupervised pre-training techniques, wav2vec 2.0, HuBERT, and data2vec. When compared to the baseline methods, the proposed CFDRN with the three pre-training techniques achieved 13.73%similar to 16.23% and 4.50%similar to 13.20% word error rate reductions on the TORGO and UASpeech datasets, respectively. Furthermore, this study clarified several major factors affecting dysarthric ASR performance.

关键词： Adaptation automatic speech recognition dysarthria

来源：评论

学校读者我要写书评

暂无评论

H2LMER: A Cross Frame-Rate Representation Alignment Framework for Micro-expression Recognition 7th

H2LMER: A Cross Frame-Rate Representation Alignment Framewor...

引用

7th Chinese Conference on Pattern Recognition and computer Vision

作者： Mao, Xinglong Liu, Shifeng Zhao, Sirui Zhang, Yiming Wang, Hao Xu, Tong Chen, Enhong Univ Sci & Technol China Sch Artificial Intelligence & Data Sci Hefei Peoples R China Univ Sci & Technol China Sch Comp Sci & Technol Hefei Peoples R China State Key Lab Cognit Intelligence Hefei Peoples R China

ISBN: (纸本)9789819787944;9789819787951

Micro-expression (ME) is a spontaneous psychological stress response characterized by subtle and fleeting facial movements, which can reflect an individual's genuine emotion. Recognizing ME accurately can provide crucial technical support for fields such as lie detection, hence attracting increasing attention from researchers in psychology and artificial intelligence. However, current automated ME recognition (MER) methods typically require ME data captured by high-speed cameras to achieve good performance, which is challenging to perform effectively in the wild with low frame-rate. To this end, we propose a novel cross frame-rate representation alignment MER framework, termed H2LMER, to enhance the low frame-rate ME feature learning. Specifically, we first customize a progressive training strategy and a cross frame-rate similarity loss to align the representation of high and low frame-rate ME data in emotion feature space. Additionally, to eliminate the interference of facial identity, we introduce a triplet loss for discriminative emotion feature learning. Afterwards, we transfer the effectively pretrained feature extractor to target low frame-rate ME datasets by partial layers finetuning. Finally, extensive experiments on CAS(ME)(3) PART A and C datasets have varified the effectiveness of our H2LMER, which outperforms single-stage baselines and existing state-of-the-art MER methods.

关键词： Micro-expression recognition Neural networks Cross frame-rate learning Representation alignment

来源：评论

学校读者我要写书评

暂无评论

An efficient continuous control perspective for reinforcement-learning-based sequential recommendation

引用

KNOWLEDGE-BASED SYSTEMS 2025年 312卷

作者： Wang, Jun Wu, Likang Liu, Qi Yang, Yu City Univ Hong Kong Sch Data Sci Kowloon Hong Kong Peoples R China Tianjin Univ Coll Management & Econ Tianjin Peoples R China Univ Sci & Technol China Anhui Prov Key Lab Big Data Anal & Applicat Hefei Anhui Peoples R China State Key Lab Cognit Intelligence Hefei Anhui Peoples R China

Sequential recommendation, where user preference is dynamically inferred from sequential historical behaviors, is a critical task in recommender systems (RSs). To further optimize long-term user engagement, offline reinforcement-learning-based RSs have become a mainstream technique as they provide an additional advantage in avoiding global explorations that may harm online users' experiences. However, previous studies mainly focus on discrete action and policy spaces, which might have difficulties in handling dramatically growing items efficiently. To mitigate this issue, in this paper, we aim to design an algorithmic framework applicable to continuous policies. To facilitate the control in the low-dimensional but dense user preference space, we propose an Efficient Continuous Control framework (ECoC). Based on a statistically tested assumption, we first propose the novel unified action representation abstracted from normalized user and item spaces. Then, we develop the corresponding policy evaluation and policy improvement procedures. During this process, strategic exploration and directional control in terms of unified actions are carefully designed and crucial to final recommendation decisions. Moreover, beneficial from unified actions, the conservatism regularization for policies and value functions are combined and perfectly compatible with the continuous framework. The resulting dual regularization ensures the successful offline training of RL-based recommendation policies. Finally, we conduct extensive experiments to validate the effectiveness of our framework. The results show that compared to the discrete baselines, our ECoC is trained far more efficiently. Meanwhile, the final policies outperform baselines in both capturing the offline data and gaining long-term rewards.

关键词： Sequential recommendation Offline reinforcement learning User preference modeling

来源：评论

学校读者我要写书评

暂无评论

Alignment-Free RGBT Salient Object Detection: Semantics-Guided Asymmetric Correlation Network and a Unified Benchmark

引用

IEEE TRANSACTIONS ON MULTIMEDIA 2024年 26卷 10692-10707页

作者： Wang, Kunpeng Lin, Danying Li, Chenglong Tu, Zhengzheng Luo, Bin Anhui Univ Sch Comp Sci & Technol Informat Mat & Intelligent Sensing Lab Anhui Prov Anhui Prov Key Lab Multimodal Cognit Computat Hefei 230601 Peoples R China Anhui Univ Sch Artificial Intelligence Informat Mat & Intelligent Sensing Lab Anhui Prov Anhui Prov Key Lab Secur Artificial Intelligence Hefei 230601 Peoples R China

RGB and Thermal (RGBT) Salient Object Detection (SOD) aims to achieve high-quality saliency prediction by exploiting the complementary information of visible and thermal image pairs, which are initially captured in an unaligned manner. However, existing methods are tailored for manually aligned image pairs, which are labor-intensive, and directly applying these methods to original unaligned image pairs could significantly degrade their performance. In this paper, we make the first attempt to address RGBT SOD for initially captured RGB and thermal image pairs without manual alignment. Specifically, we propose a Semantics-guided Asymmetric Correlation Network (SACNet) that consists of two novel components: 1) an asymmetric correlation module utilizing semantics-guided attention to model cross-modal correlations specific to unaligned salient regions;2) an associated feature sampling module to sample relevant thermal features according to the corresponding RGB features for multi-modal feature integration. In addition, we construct a unified benchmark dataset called UVT2000, containing 2000 RGB and thermal image pairs directly captured from various real-world scenes without any alignment, to facilitate research on alignment-free RGBT SOD. Extensive experiments on both aligned and unaligned datasets demonstrate the effectiveness and superior performance of our method.

关键词： Correlation Feature extraction Manuals Object detection Convolution Benchmark testing Transformers Alignment-free associated feature sampling module asymmetric correlation module RGBT salient object detection

来源：评论

学校读者我要写书评

暂无评论

TransO: a knowledge-driven representation learning method with ontology information constraints

引用

WORLD WIDE WEB-INTERNET AND WEB INFORMATION SYSTEMS 2023年第1期26卷 297-319页

作者： Li, Zhao Liu, Xin Wang, Xin Liu, Pengkai Shen, Yuxin Tianjin Univ Coll Intelligence & Comp Tianjin Peoples R China Tianjin Key Lab Cognit Comp & Applicat Tianjin Peoples R China Bank Shizuishan Financial Big Data Lab Yinchuan Ningxia Peoples R China Tianjin Univ Tianjin Int Engn Inst Tianjin Peoples R China

Representation learning techniques for knowledge graphs (KGs) are crucial for constructing knowledge-driven decisions in complex network data application scenarios. Most existing methods focus mainly on structured information, ignoring the important value of rich ontology information constraints and complements, however, ontology information is the key for building knowledge-driven decision-making processes. In this paper, we propose a novel ontology information constrained knowledge representation learning model, TransO, which can efficiently model relations explicitly and seamlessly incorporate rich ontology information to improve model performance and maintain low model complexity. Moreover, specific constraint strategies are proposed for entity types, relations, and hierarchical information to effectively implement reasoning and completion of KGs and construct knowledge-driven decisions that are more consistent with the logic of human knowledge in complex network applications. The experimental tasks of link prediction and triple classification are performed on two public datasets. The experimental results demonstrate the effectiveness of our proposed method with better performance than state-of-the-art methods.

关键词： Knowledge graph Representation learning Ontology information

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：