检索结果-内蒙古大学图书馆

Under-determined audio source separation using the convolutive narrowband approximation and flexible ℓ p regularizer

引用

Applied Acoustics 2025年 240卷

作者： Junjie Yang Liu Yang Yi Guo Yuan Xie Shengli Xie School of Automation Guangdong University of Technology Guangdong Key Laboratory of IoT Information Technology (GDUT) and Key Laboratory of Intelligent Information Processing and System Integration of IoT (GDUT) Ministry of Education Guangzhou 510006 China School of Computer Science and Cyber Engineering Guangzhou University Guangzhou 510006 China School of Computer Data and Mathematical Sciences Western Sydney University Parramatta 2150 NSW Australia Mechanical and Electrical Engineering Guangzhou University and Key Laboratory of Intelligent Detection and The Internet of Things in Manufacturing (GDUT) Ministry of Education Guangzhou 510006 China

Convolutive narrowband approximation (CNA) model is widely utilized in audio source separation, particularly in strongly reverberant scenarios. However, in under-determined audio source separation, the CNA model often faces a serious ill-conditioned inverse filtering problem due to its mixing matrices containing columns that approach zero. To mitigate this issue, a ℓ p regularizer-based approach is proposed in this paper, assuming mixing filters of CNA model are known or pre-estimated. First, the STFT mixture components of all frames at each frequency bin are concatenated into a long vector, and the CNA system is accordingly expanded to a block-circulant structured linear mixing model. Next, a maximum likelihood estimation, constrained by the proposed mixing model, is introduced to exploit the sparsity of STFT source components with the ℓ p regularizer, where these components are assumed to independently follow super-Gaussian distributions. Finally, an augmented Lagrange multiplier with efficient iterative strategy is developed to search for the suitable sparse solution. The proposed strategy with regularizer p can be flexibly selected in a range of ( 0 , 1 ] to achieve robustness and accuracy in performance. Experimental results in various under-determined cases demonstrate the superior performance of the proposed algorithm over state-of-the-art approaches.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Domain Adaptive Object Detection Based on Channel Recalibration and Ancillary Net 4

Unsupervised Domain Adaptive Object Detection Based on Chann...

引用

4th IEEE Advanced information Management, Communicates, Electronic and Automation Control Conference, IMCEC 2021

作者： Wang, Qin Jiang, Siyu Ma, Le Zheng, Chunyun Su, Xiaolei Shi, Ruihao South China University of Technology School of Mechanical and Automotive Engineering Guangzhou China Guangdong University of Foreign Studies Guangzhou Key Laboratory of Multilingual Intelligent Processing School of Information Science and Technology Guangzhou China Guangzhou City Institute of Technology Guangzhou China Guangzhou Automobile Group Co. Ltd Automotive Engineering Research Institute Guangzhou China Baoneng New Energy Automobile Group Co. Ltd Shenzhen China

ISBN: (纸本)9781728185347

In the field of object detection, it is generally hoped that the model trained in the label-rich environment can be well applied to other various and label-agnostic environment. However, due to the existence of domain discrepancy between the training environment and the application environment, the performance of conventional object detection models inevitably encounters a sharp drop. Aiming at the limitations of the object detection application environment and the time-consuming of image labeling, this paper proposes an unsupervised domain adaptive object detection model based on Faster R-CNN. We integrate SENets into the feature extraction network to capture channel-wise correlations between features and strengthen the representations produced by CNNs, and add ancillary nets at the image level and the instance level respectively to prevent data distribution distortion caused by adversarial training. The experiments show that our proposed approach not only enhances the robustness of object detection under different environmental conditions, but also improves the detection accuracy of objects. © 2021 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Group-wise inhibition based feature regularization for robust classification

arXiv

引用

arXiv 2021年

作者： Liu, Haozhe Wu, Haoqian Xie, Weicheng Liu, Feng Shen, Linlin 1Computer Vision Institute College of Computer Science and Software Engineering 2SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society 3National Engineering Laboratory for Big Data System Computing Technology 4Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen 518060 China

The convolutional neural network (CNN) is vulnerable to degraded images with even very small variations (e.g. corrupted and adversarial samples). One of the possible reasons is that CNN pays more attention to the most discriminative regions, but ignores the auxiliary features when learning, leading to the lack of feature diversity for final judgment. In our method, we propose to dynamically suppress significant activation values of CNN by group-wise inhibition, but not fixedly or randomly handle them when training. The feature maps with different activation distribution are then processed separately to take the feature independence into account. CNN is finally guided to learn richer discriminative features hierarchically for robust classification according to the proposed regularization. Our method is comprehensively evaluated under multiple settings, including classification against corruptions, adversarial attacks and low data regime. Extensive experimental results show that the proposed method can achieve significant improvements in terms of both robustness and generalization performances, when compared with the state-of-the-art methods. Code is available at https://***/LinusWu/TENET_Training. Copyright © 2021, The Authors. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Rate Splitting Multiple Access for Next Generation Cognitive Radio Enabled LEO Satellite Networks

arXiv

引用

arXiv 2022年

作者： Khan, Wali Ullah Ali, Zain Lagunas, Eva Mahmood, Asad Asif, Muhammad Ihsan, Asim Chatzinotas, Symeon Ottersten, Björn Dobre, Octavia A. University of Luxembourg Luxembourg City1855 Luxembourg Department of Electrical and Computer Engineering University of California Santa Cruz United States Guangdong Key Laboratory of Intelligent Information Processing College of Electronics and Information Engineering Shenzhen Univerisyt Guangdong Shenzhen China The School of Computer Science and Electronic Engineering Bangor University BangorLL57 1UT United Kingdom The Department of Electrical and Computer Engineering Faculty of Engineering and Applied Science Memorial University St. John’sNLA1B 3X9 Canada

Low Earth Orbit (LEO) satellite communication (SatCom) has drawn particular attention recently due to its high data rate services and low round-trip latency. It has low launching and manufacturing costs than Medium Earth Orbit (MEO) and Geostationary Earth Orbit (GEO) satellites. Moreover, LEO SatCom has the potential to provide global coverage with a high-speed data rate and low transmission latency. However, the spectrum scarcity might be one of the challenges in the growth of LEO satellites, impacting severe restrictions on developing ground-space integrated networks. To address this issue, cognitive radio and rate splitting multiple access (RSMA) are the two emerging technologies for high spectral efficiency and massive connectivity. This paper proposes a cognitive radio enabled LEO SatCom using RSMA radio access technique with the coexistence of GEO SatCom network. In particular, this work aims to maximize the sum rate of LEO SatCom by simultaneously optimizing the power budget over different beams, RSMA power allocation for users over each beam, and subcarrier user assignment while restricting the interference temperature to GEO SatCom. The problem of sum rate maximization is formulated as non-convex, where the global optimal solution is challenging to obtain. Thus, an efficient solution can be obtained in three steps: first we employ a successive convex approximation technique to reduce the complexity and make the problem more tractable. Second, for any given resource block user assignment, we adopt Karush–Kuhn–Tucker (KKT) conditions to calculate the transmit power over different beams and RSMA power allocation of users over each beam. Third, using the allocated power, we design an efficient algorithm based on the greedy approach for resource block user assignment. For comparison, we propose two suboptimal schemes with fixed power allocation over different beams and random resource block user assignment as the benchmark. Numerical results provided in this wo

关键词： Spectrum efficiency

来源：评论

学校读者我要写书评

暂无评论

AdAUC: End-to-end Adversarial AUC Optimization Against Long-tail Problems

arXiv

引用

arXiv 2022年

作者： Hou, Wenzheng Xu, Qianqian Yang, Zhiyong Bao, Shilong He, Yuan Huang, Qingming Key Laboratory of Intelligent Information Processing Institute of Computing Technology CAS Beijing China School of Computer Science and Technology University of Chinese Academy of Sciences Beijing China State Key Laboratory of Information Security Institute of Information Engineering CAS Beijing China School of Cyber Security University of Chinese Academy of Sciences Beijing China Alibaba Group Beijing China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing China Artificial Intelligence Research Center Peng Cheng Laboratory Shenzhen China

It is well-known that deep learning models are vulnerable to adversarial examples. Existing studies of adversarial training have made great progress against this challenge. As a typical trait, they often assume that the class distribution is overall balanced. However, long-tail datasets are ubiquitous in a wide spectrum of applications, where the amount of head class instances is larger than the tail classes. Under such a scenario, AUC is a much more reasonable metric than accuracy since it is insensitive toward class distribution. Motivated by this, we present an early trial to explore adversarial training methods to optimize AUC. The main challenge lies in that the positive and negative examples are tightly coupled in the objective function. As a direct result, one cannot generate adversarial examples without a full scan of the dataset. To address this issue, based on a concavity regularization scheme, we reformulate the AUC optimization problem as a saddle point problem, where the objective becomes an instance-wise function. This leads to an end-to-end training protocol. Furthermore, we provide a convergence guarantee of the proposed algorithm. Our analysis differs from the existing studies since the algorithm is asked to generate adversarial examples by calculating the gradient of a min-max problem. Finally, the extensive experimental results show the performance and robustness of our algorithm in three long-tail datasets. © 2022, CC BY.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Holographic data compression with JPEG standard and deep learning

Holographic data compression with JPEG standard and deep lea...

引用

Digital Holography and Three-Dimensional Imaging, DH 2019

作者： Gao, Yang Jiao, Shuming Jin, Zhi Nanophotonics Research Centre College of Optoelectronic Engineering Shenzhen University Shenzhen Guangdong China Guangdong Key Laboratory of Intelligent Information Processing College of Information Engineering Shenzhen University Shenzhen Guangdong China

We propose a phase-only hologram compression scheme based on "JPEG standard + deep learning". Our scheme is compatible with universal JPEG standard and the compression artifacts are effectively restored by a... 详细信息

ISBN: (纸本)9781943580590

关键词： Holograms

来源：评论

学校读者我要写书评

暂无评论

Multi-scale feature fusion for breast tumor grading using dual-adaptive attention mechanisms

引用

Pattern Recognition 2026年 169卷

作者： Israr Hussain Zhihui Lai Heng Kong Fazil Hussain College of Computer Science and Software Engineering Shenzhen University Shenzhen 518060 China Guangdong Provincial Key Laboratory of Intelligent Information Processing Shenzhen 518060 China Department of Thyroid and Breast Surgery BaoAn Central Hospital of Shenzhen Shenzhen 518102 China Department of Medicine Provincial Head Quarter (PHQ) Hospital Gilgit 15711 Pakistan

Breast cancer is the most common malignant tumor in women worldwide. Early detection and timely treatment are crucial for improving survival rates and ensuring effective treatment planning. While existing methods primarily focus on classifying benign versus malignant tumors, accurate tumor grading remains underexplored. The scarcity of high-quality annotated datasets for advanced tumor grades, combined with their inherent similarity, limits model robustness and precise grading. To address these challenges, we propose a novel framework for breast tumor grading diagnosis that leverages multi-scale feature fusion with dual-adaptive attention mechanisms. This framework enhances grading performance by capturing both local and global information. It integrates information across multiple scales, improves feature representation, and enables the detection of subtle differences between grades. The framework consists of three primary modules: deep hybrid feature extraction, multi-scale feature fusion, and dual-adaptive attention mechanisms. Together, these modules enhance feature representation, capture salient information, and highlight discriminative features, ultimately improving grading accuracy. In addition, we constructed two distinct image datasets consisting of 17,058 ultrasound images and 1571 MRI images for breast tumor grading. Experimental results demonstrated the effectiveness of our framework, achieving a test accuracy of 0.89, a macro-average F1-score of 0.82, and a micro-average precision of 0.93, outperforming existing state-of-the-art methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MaxMatch: Semi-Supervised Learning with Worst-Case Consistency

arXiv

引用

arXiv 2022年

作者： Jiang, Yangbangyan Li, Xiaodan Chen, Yuefeng He, Yuan Xu, Qianqian Yang, Zhiyong Cao, Xiaochun Huang, Qingming State Key Laboratory of Information Security Institute of Information Engineering Chinese Academy of Sciences Beijing100093 China School of Cyber Security University of Chinese Academy of Sciences Beijing100049 China Security Department of Alibaba Group Hangzhou311121 China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China School of Computer Science and Technology University of Chinese Academy of Sciences Beijing101408 China School of Cyber Science and Technology Shenzhen Campus Sun Yat-sen University Shenzhen518107 China University of Chinese Academy of Sciences Beijing101408 China Peng Cheng Laboratory Shenzhen518055 China

In recent years, great progress has been made to incorporate unlabeled data to overcome the inefficiently supervised problem via semi-supervised learning (SSL). Most state-of-the-art models are based on the idea of pursuing consistent model predictions over unlabeled data toward the input noise, which is called consistency regularization. Nonetheless, there is a lack of theoretical insights into the reason behind its success. To bridge the gap between theoretical and practical results, we propose a worst-case consistency regularization technique for SSL in this paper. Specifically, we first present a generalization bound for SSL consisting of the empirical loss terms observed on labeled and unlabeled training data separately. Motivated by this bound, we derive an SSL objective that minimizes the largest inconsistency between an original unlabeled sample and its multiple augmented variants. We then provide a simple but effective algorithm to solve the proposed minimax problem, and theoretically prove that it converges to a stationary point. Experiments on five popular benchmark datasets validate the effectiveness of our proposed method. Copyright © 2022, The Authors. All rights reserved.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

GRATIS: Deep Learning Graph Representation with Task-specific Topology and Multi-dimensional Edge Features

arXiv

引用

arXiv 2022年

作者： Song, Siyang Song, Yuxin Luo, Cheng Song, Zhiyuan Kuzucu, Selim Jia, Xi Guo, Zhijiang Xie, Weicheng Shen, Linlin Gunes, Hatice The Department of Computer Science and Technology University of Cambridge CambridgeCB3 0FT United Kingdom Baidu Inc Beijing100193 China Computer Vision Institute School of Computer Science & Software Engineering Shenzhen Institute of Artificial Intelligence and Robotics for Society Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen518060 China Department of Computer Engineering Middle East Technical University Ankara Turkey School of Computer Science University of Birmingham Birmingham United Kingdom

Graph is powerful for representing various types of real-world data. The topology (edges’ presence) and edges’ features of a graph decides the message passing mechanism among vertices within the graph. While most existing approaches only manually define a single-value edge to describe the connectivity or strength of association between a pair of vertices, task-specific and crucial relationship cues may be disregarded by such manually defined topology and single-value edge features. In this paper, we propose the first general graph representation learning framework (called GRATIS) which can generate a strong graph representation with a task-specific topology and task-specific multi-dimensional edge features from any arbitrary input (i.e., graph or non-graph data). To learn each edge’s presence and multi-dimensional feature, our framework takes both of the corresponding vertices pair and their global contextual information into consideration, enabling the generated graph representation to have a globally optimal message passing mechanism for different down-stream tasks. The principled investigation results achieved for various graph analysis tasks (e.g., graph classification, vertex classification, and link prediction) on 11 graph and non-graph datasets show that our GRATIS can not only largely enhance pre-defined graphs but also learns a strong graph representation for non-graph data, with clear performance improvements on all tasks. In particular, the learned topology and multi-dimensional edge features provide complementary task-related cues for graph analysis tasks. Our framework is effective, robust and flexible, and is a plug-and-play module that can be combined with different backbones and Graph Neural Networks (GNNs) to generate a task-specific graph representation from various graph and non-graph data. Our code is made publicly available at https://***/SSYSteve/Learning-Graph-Representation-with-Task-specific-Topology-and-Multi-dimensional-Edge-Feature

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Radar Emitter Signal Detection with Convolutional Neural Network 11

Radar Emitter Signal Detection with Convolutional Neural Net...

引用

11th IEEE International Conference on Advanced Infocomm Technology, ICAIT 2019

作者： Liu, Zhenrong Shi, Yankun Zeng, Yuan Gong, Yi Southern University of Science and Technology Department of Electrical and Electronic Engineering Shenzhen11849160 China Southern University of Science and Technology Academy for Advanced Interdisciplinary Studies Shenzhen China Southern University of Science and Technology Shenzhen Engineering Laboratory of Intelligent Information Processing for IoT Shenzhen China

ISBN: (纸本)9781728147789

In this paper, we propose a deep convolutional neural network (CNN) based automatic detection algorithm for recognizing radar emitter signals. The algorithm leverages on the structure estimation power of deep CNN and the capability of time-frequency image processing for radio signal representation. We transform raw radio signals into time-frequency image using the Choi-Williams distribution function. We compare the proposed method with Belief Propagation (BP) and Support Vector Machine (SVM) based methods in terms of recognition accuracy versus signal-to-noise-ratio. The experiments demonstrate that the proposed CNN network with time-frequency image processing achieves very competitive results on the testing datasets. © 2019 IEEE.

关键词： Signal detection

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：