检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Wen Huang Bing Han Shuai Wang Zhengyang Chen Yanmin Qian AI Institute Department of Computer Science and Engineering Auditory Cognition and Computational Acoustics Lab MoE Key Lab of Artificial Intelligence Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

Speaker verification encounters significant challenges when confronted with diverse domain data, often resulting in performance degradation due to domain mismatch. To enhance performance in cross-domain scenarios, we introduce the Domain Adapter, an adaptable module designed for specific domains. This module learns and integrates domain-specific information with speaker-related data, mitigating domain-related variations and promoting convergence of utterance embeddings from the same speaker across diverse domains. It offers configurability across multiple levels and is adaptable to various backbone architectures. Our proposed module substantially enhances cross-domain performance with minimal parameter increments while effectively generalizing to previously unseen domains. In our experiments, we present results on the 3D-Speaker dataset, which provides acoustically-relevant attributes crucial for domain categorization and the subsequent learning of domain information. The top-performing system integrated with domain adapters achieved 10.8%, 14.8%, and 21.1% EER improvements over the baseline across three 3D-Speaker dataset trials.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Trusted Lesion-assessment Network for Interpretable Diagnosis of Coronary Artery Disease in Coronary CT Angiography 39

A Trusted Lesion-assessment Network for Interpretable Diagno...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Ma, Xinghua Fang, Xinyan Zou, Mingye Luo, Gongning Wang, Wei Wang, Kuanquan Qiu, Zhaowen Gao, Xin Li, Shuo Faculty of Computing Harbin Institute of Technology Harbin China Computational Bioscience Research Center King Abdullah University of Science and Technology Thuwal Saudi Arabia Faculty of Computing Harbin Institute of Technology Shenzhen China College of Computer and Control Engineering Northeast Forestry University Harbin China Department of Computer and Data Science Case Western Reserve University Cleveland United States Department of Biomedical Engineering Case Western Reserve University Cleveland United States

ISBN: (纸本)157735897X

Coronary Artery Disease (CAD) poses a significant threat to cardiovascular patients worldwide, underscoring the critical importance of automated CAD diagnostic technologies in clinical practice. Previous technologies for lesion assessment in Coronary CT Angiography (CCTA) images have been insufficient in terms of interpretability, resulting in solutions that lack clinical reliability in both network architecture and prediction outcomes, even when diagnoses are accurate. To address the limitation of interpretability, we introduce the Trusted Lesion-Assessment Network (TLA-Net), which provides a clinically reliable solution for multi-view CAD diagnosis: (1) The causality-informed evidence collection constructs a causal graph for the diagnostic process and implements causal interventions, preventing confounders’ interference and enhancing the transparency of the network architecture. (2) The clinically-aligned uncertainty integration hierarchically combines Dirichlet distributions from various views based on clinical priors, offering confidence coefficients for prediction outcomes that align with physicians’ image analysis procedures. Experimental results on a dataset of 2,618 lesions demonstrate that TLA-Net, supported by its interpretable methodological design, exhibits superior performance with outstanding generalization, domain adaptability, and robustness. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Advanced Fault Detection in Photovoltaic Panels Using Enhanced U-Net Architectures

SSRN

引用

SSRN 2025年

作者： Awedat, Khalfalla Comert, Gurcan Ayad, Mustafa Mrebit, Abdulmajid Computer Information Technology Department SUNY Morrisville College 80 Eaton St MorrisvilleNY13408 United States Computational Data Science and Engineering Department North Carolina A&T State University 1601 E. Market Street GreensboroNC27411 United States Electrical and Computer Engineering Department SUNY Oswego 430 Shineman Center OswegoNY13126 United States School of Art and Sciences Computer Science Physics & Engineering Department Benedict College 1600 Harden Street ColumbiaSC29204 United States

Fault detection in photovoltaic (PV) panels using thermal images remains a significant challenge due to the com-plexity of thermal patterns, environmental noise, and the subtle nature of anomalies. This paper introduces anadvanced deep learning framework that enhances the U-Net architecture by integrating Residual Blocks, AtrousSpatial Pyramid Pooling (ASPP), and Attention Mechanisms. These enhancements collectively improve featureextraction, contextual understanding, and fault localization, addressing the limitations of traditional segmentationapproaches and reducing false positives. Extensive experiments demonstrate that the proposed method significantlyoutperforms all benchmarked algorithms across key segmentation metrics, including standard U-Net, U-Net withASPP, and DeepLabV3+. Compared to standard U-Net, the proposed model achieves more than a 29% increasein F1-score and a 62% improvement in Intersection over Union (IoU) while reducing segmentation loss by 71%.Its ability to accurately detect faults under challenging conditions establishes the framework as a state-of-the-artsolution for real-time PV monitoring. These results demonstrate the effectiveness of the proposed approach in ad-dressing the challenges of PV fault detection, offering a practical and reliable solution for ensuring the operationalperformance of renewable energy systems. © 2025, The Authors. All rights reserved.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Optimizing Forwarding Strategies in Sparsely Delay-Tolerant Networks with Short Contact Durations

Optimizing Forwarding Strategies in Sparsely Delay-Tolerant ...

引用

International Conference on Computing and Networking Technology (ICCNT)

作者： Yadesa Bedasa Nemera Leta Tesfaye Jule S. Sivakumar J Pavithra A. Sampathkumar Ravi Rastogi Department of Information Technology Dambi Dollo Univeristy Ethiopia Department of Physics College of Natural and Computational Science Dambi Dollo University Dambi Dollo Ethiopia Department of CSE Nehru Institute of Engineering and Technology Coimbatore India Department of Artificial intelligence and data science Dhanalakshmi Srinivasan College of Engineering Coimbatore India Department of Computer Science Dambi Dollo Univeristy Ethiopia Department of CSE Koneru Lakshmaiah Education Foundation Guntur AP India

ISBN: (数字)9798350370249

ISBN: (纸本)9798350370270

Delay-tolerant networks (DTNs) are specific settings that experience either instability or the absence of a complete path between the source and destination nodes. The contact lifetime between pairs of network nodes is constrained by the mobility of nodes and the restricted bandwidth of peer- to-peer wireless transmission. In sparsely delay-tolerant networks, the re-establishment of contact between pairs of nodes is significantly delayed beyond the duration of the initial contact. Hence, the need for careful thought in designing a buffer control forwarding policy arises due to the limited contact period. Although forwarding algorithms have been designed extensively, much research has not been done on forwarding when there is a short contact time. When the message exchange between the pairs of network nodes is not completed during the contract period, they must wait until they are within communication range once more. To tackle the issue, we have suggested an innovative forwarding policy in buffer management that considers the duration of interaction. To assess the effectiveness of the proposed effort, we utilized the Opportunistic Network Environment (ONE) networking simulator. The suggested approach is combined with the MaxProp routing protocol buffer control forwarding policy and evaluated using two distinct metrics: bundle delivery rate and average delivery delay. In addition, we evaluated the performance of the suggested work by considering the number of nodes, buffer size, and simulation time. Across all the trials conducted in this thesis, the simulation results consistently demonstrate that our suggested method beats the existing forwarding strategy.

关键词： Wireless communication Measurement Simulation Receivers Programming Routing protocols Product delivery Peer-to-peer computing Delays Contracts

来源：评论

学校读者我要写书评

暂无评论

Multilevel Monte Carlo for a class of Partially Observed Processes in Neuroscience

arXiv

引用

arXiv 2023年

作者： Maama, Mohamed Jasra, Ajay Kamatani, Kengo Applied Mathematics and Computational Science Program Computer Electrical and Mathematical Sciences and Engineering Division King Abdullah University of Science and Technology Thuwal23955-6900 Saudi Arabia School of Data Science The Chinese University of Hong Kong CN Shenzhen China Institute of Statistical Mathematics Tokyo190-0014 Japan

In this paper we consider Bayesian parameter inference associated to a class of partially observed stochastic differential equations (SDE) driven by jump processes. Such type of models can be routinely found in applications, of which we focus upon the case of neuroscience. The data are assumed to be observed regularly in time and driven by the SDE model with unknown parameters. In practice the SDE may not have an analytically tractable solution and this leads naturally to a time-discretization. We adapt the multilevel Markov chain Monte Carlo method of [15], which works with a hierarchy of time discretizations and show empirically and theoretically that this is preferable to using one single time discretization. The improvement is in terms of the computational cost needed to obtain a pre-specified numerical error. Our approach is illustrated on models applied to examples of Bayesian inference problems using both simulated and real observed data from neuroscience. Copyright © 2023, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

A consensus model for linguistic group decision making considering emotion analysis

引用

Procedia computer science 2023年 221卷 1235-1241页

作者： Mengru Xu Fuying Jing Xiangrui Chao Enrique Herrera-viedma School of Management Science and Engineering Chongqing Technology and Business University Chongqing 400067 China National Research Base of Intelligent Manufacturing Service Chongqing Technology and Business University Chongqing 400067 China School of Business Sichuan University Chengdu 610065 China Andalusian Research Institute in Data Science and Computational Intelligence Department of Computer Science and Artificial Intelligence University of Granada Granada 18071 Spain

In a problem of linguistic group decision-making (GDM), various emotional characteristics of decision-makers’ (DMs) behind their preference series, which is defined as emotion factor, can impact consensus reaching process. As words mean different things to different people, this study comprehensively proposes a novel consensus model with emotion factors based on personalized individual semantics. An improved grey clustering method facilitates to obtain emotion factors. Finally, the proposal model can provide more accurate judgments for moderator's management and further promote consensus.

关键词： Emotion factors linguistic group decision-making personalized individual semantics grey clustering

来源：评论

学校读者我要写书评

暂无评论

Autonomous Decision-Making in Cyber-Physical Systems Using Double Deep Q-Learning and Multi-Agent Collaboration

Autonomous Decision-Making in Cyber-Physical Systems Using D...

引用

Advancement in Computation & computer Technologies (InCACCT), International Conference on

作者： Divyabarathi P Anbarasan R. Vasanthi R B. Leena Akshya J M. Sundarrajan Mani Deepak Choudhry Department of Artificial Intelligence and Data Science Bannari Amman Institute of Technology Erode Tamil Nadu India Department of Information Technology KGiSL Institute of Technology Coimbatore Tamil Nadu India Department of Computer Science and Engineering Bharathidasan Engineering College Thirupathur Tamil Nadu India Department of Electronics and Communication Engineering KGiSL Institute of Technology Coimbatore Tamil Nadu India Department of Computational Intelligence SRM Institute of Science and Technology Chennai Tamil Nadu India Department of Networking and Communications SRM Institute of Science and Technology Chennai Tamil Nadu India Department of Computing Technologies SRM Institute of Science and Technology Chennai Tamil Nadu India

ISBN: (数字)9798331543891

ISBN: (纸本)9798331543907

The autonomous decision-making capabilities of cyber-physical systems (CPS) establish key elements that optimize the management of resources while distributing tasks and maximizing system operational efficiency. Deep Q-Learning (DQN) represents one approach along with Twin Delayed Deep Deterministic Policy Gradient (TD3) and Multi-Agent Proximal Policy Optimization (PPO) that demonstrates capability in handling decision tasks. The existing models face multiple implementation challenges including inaccurate Q-value estimations weak multi-agent relationships and restricted scalability for dynamic systems. The inadequate performance of present methods for energy efficiency and latency alongside poor collaborative coordination led to the necessity of developing a new approach. The new framework DDQN-MAC takes double-deep Q-Learning and combines it with multi-agent collaboration technology to enhance decision-making abilities specific to challenging environments. The method solves overestimation bias through separate action selection and evaluation processes that use multi-agent collaboration for dynamic task distribution and resource control. The designed model demonstrates enhanced performance in multiple critical evaluation measures. Benchmarks indicate that DDQN-MAC attained a superior completion rate of 95.4% while demonstrating reduced latency performance at 12.5 ms versus the other methods DQN (18.7 ms) and Multi-Agent PPO (92.3%). The research provides fundamental groundwork to guide upcoming investigations in hybrid model development while advancing complex multi-agent system capabilities for enhanced autonomous navigational choices within variable operational environments.

关键词： Q-learning Decision making Collaboration Estimation Cyber-physical systems Dynamic scheduling Energy efficiency Resource management Vehicle dynamics Optimization

来源：评论

学校读者我要写书评

暂无评论

Day-Ahead Forecasting of Solar Irradiance & PV Power Output Through Statistical Machine Learning Methods

Day-Ahead Forecasting of Solar Irradiance & PV Power Output ...

引用

2022 Saudi Arabia Smart Grid Conference, SASG 2022

作者： Kamarianakis, Yiannis Pantazis, Yannis Kalligiannaki, Evangelia Katsaounis, Theodoros D. Kotsovos, Konstantinos Gereige, Issam Abdullah, Marwan Jamal, Aqil Tzavaras, Athanasios Foundation for Research and Technology-Hellas Institute of Applied and Computational Mathematics Data Science Group Greece University of Crete Dept. of Mathematics and Applied Mathematics Greece Saudi Aramco RandD Center Carbon Management Division Dhahran Saudi Arabia Computer Electrical and Mathematical Sciences and Engineering Division Thuwal Saudi Arabia

ISBN: (纸本)9781665475891

Energy production from solar photovoltaic (PV) plants is unpredictable, mainly due to the stochastic formation and movement of clouds or aerosol - dust particles which scatter or disperse solar radiation. Accurate forecasts of PV output are essential to Distribution and Transportation System Operators as they assist efficient solar energy trading and management of electricity grids. This work evaluates an autoregressive, computationally-light KNN-regression scheme (TSFKNN) for hourly, day-ahead forecasts of solar irradiance and energy yield of various PV technologies. The model is being tested and validated using data measured in Thuwal, Saudi Arabia. The available measured records span a 60-month period. The developed forecasting models are designed for online systems and provide increased levels of accuracy and low computational cost. Several parametric and nonparametric specifications, coupled with conventional versus outlier-robust estimation procedures are tested, in order to derive an optimal month-specific daily profile (MDP). Current results demonstrate that including intraday variability to the monthly-based irradiance models achieve improved predictive accuracy between 10% and 25% on average. © 2022 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Prototype and Instance Contrastive Learning for Unsupervised Domain Adaptation in Speaker Verification

arXiv

引用

arXiv 2024年

作者： Huang, Wen Han, Bing Chen, Zhengyang Wang, Shuai Qian, Yanmin Auditory Cognition and Computational Acoustics Lab MoE Key Lab of Artificial Intelligence AI Institute Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

Speaker verification system trained on one domain usually suffers performance degradation when applied to another domain. To address this challenge, researchers commonly use feature distribution matching-based methods in unsupervised domain adaptation scenarios where some unlabeled target domain data is available. However, these methods often have limited performance improvement and lack generalization in various mismatch situations. In this paper, we propose Prototype and Instance Contrastive Learning (PICL), a novel method for unsupervised domain adaptation in speaker verification through dual-level contrastive learning. For prototype contrastive learning, we generate pseudo labels via clustering to create dynamically updated prototype representations, aligning instances with their corresponding class or cluster prototypes. For instance contrastive learning, we minimize the distance between different views or augmentations of the same instance, ensuring robust and invariant representations resilient to variations like noise. This dual-level approach provides both high-level and low-level supervision, leading to improved generalization and robustness of the speaker verification model. Unlike previous studies that only evaluated mismatches in one situation, we have conducted relevant explorations on various datasets and achieved state-of-the-art performance currently, which also proves the generalization of our method. Copyright © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

EvoSampling: A Granular Ball-based Evolutionary Hybrid Sampling with Knowledge Transfer for Imbalanced Learning

arXiv

引用

arXiv 2024年

作者： Pei, Wenbin Dai, Ruohao Xue, Bing Zhang, Mengjie Zhang, Qiang Cheung, Yiu-Ming Xia, Shuyin The School of Computer Science and Technology Dalian University of Technology Dalian116024 China Ministry of Education Dalian116024 China The School of Engineering and Computer Science Victoria University of Wellington PO Box 600 Wellington6140 New Zealand The Department of Computer Science Hong Kong Baptist University Hong Kong The Chongqing Key Laboratory of Computational Intelligence Key Laboratory of Big Data Intelligent Computing Key Laboratory of Cyberspace Big Data Intelligent Security Ministry of Education Chongqing University of Posts and Telecommunications Chongqing400065 China

Class imbalance would lead to biased classifiers that favor the majority class and disadvantage the minority class. Unfortunately, from a practical perspective, the minority class is of importance in many real-life applications. Hybrid sampling methods address this by oversampling the minority class to increase the number of its instances, followed by undersampling to remove low-quality instances. However, most existing sampling methods face difficulties in generating diverse high-quality instances and often fail to remove noise or low-quality instances on a larger scale effectively. This paper therefore proposes an evolutionary multi-granularity hybrid sampling method, called EvoSampling. During the oversampling process, genetic programming (GP) is used with multi-task learning to effectively and efficiently generate diverse high-quality instances. During the undersampling process, we develop a granular ball-based undersampling method that removes noise in a multi-granular fashion, thereby enhancing data quality. Experiments on 20 imbalanced datasets demonstrate that EvoSampling effectively enhances the performance of various classification algorithms by providing better datasets than existing sampling methods. Besides, ablation studies further indicate that allowing knowledge transfer accelerates the GP's evolutionary learning process. Copyright © 2024, The Authors. All rights reserved.

关键词：

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：