检索结果-内蒙古大学图书馆

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 1208-1219页

作者： Ying Hu Qin Yang Wenbing Wei Li Lin Liang He Zhijian Ou Wenzhong Yang School of Computer Science and Technology Key Laboratory of Signal Detection and Processing Xinjiang University Urumqi China Xinjiang Institute of Electronics Research Shares Company Ltd. Urumqi China Department of Electrical Engineering Tsinghua University Beijing China Beijing National Research Center for information Science and Technology Department of Electronics Tsinghua University Beijing China

Currently, deep learning-based speech enhancement methods generally focus on target speech extraction while neglecting modeling the other sound sources in the mixture. These methods still can't distinguish the target speech from the interference well. In this paper, we present a monaural speech enhancement network via Modeling the Noise (MN-Net), which includes a shared Encoder and three separate Decoders for parallel modeling the magnitude and phase spectrogram of target speech, and the complex spectrogram of noise. Specifically, we propose a Multi-Branch Feature Extractor (MBFE) module to capture the richer contextual information in mixture, and a Spatial Reconstruction Unit (SRU) to remove the redundancy from extracted features. We compared our proposed MN-Net with 18 classical speech enhancement methods on the VoiceBank+DEMAND dataset, and with 9 ones on DNS-Challenge dataset for denoising task, and with 7 ones on the WHAMR! dataset for simultaneous denoising & de-reverberation task. Our proposed MBFE module was applied to two classical speech enhancement methods, DB-AIAT and CMGAN, replacing their DenseBlocks module. The results demonstrate that applying the MBFE module can boost their performances while keeping smaller model size. A series of visualization analysis intuitively verify that modeling the noise can enable the network to distinguish the target speech from noise and other interference more accurately.

关键词： Noise Feature extraction Spectrogram Speech enhancement Decoding Kernel Transformers Noise reduction Data mining Training

来源：评论

学校读者我要写书评

暂无评论

FDAM: Filter-Dedicated Approximate Multiplier Design for Real-Time CNN Acceleration

引用

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年

作者： Nguyen, Kihwan Kim Xuan Truong Jeon, Dongsuk Lee, Hyuk-Jae Korea Republic of program Department of Electrical Engineering Seoul08826 Korea Republic of Inter-University Semiconductor Research Center Seoul National University Graduate School of Convergence Science and Technology Research Institute for Convergence Science Seoul08826 Korea Republic of Department of Electrical and Computer Engineering Seoul08826 Korea Republic of

Video super-resolution (VSR) is widely used in various high-definition applications, such as HDTVs and smartphones, requiring a dedicated upscaling technique for realtime full-HD generation. To reduce on-chip buffers for largesize output feature maps, a streaming VSR accelerator may employ an output stationary dataflow, leading to a large energy consumption caused by frequent filter switching. To mitigate this, we introduce a new filter-dedicated multiplier design for realtime VSR acceleration. We replace costly multipliers with adders, shifters, and multiplexers (MUXes), referred to as unified multiple constant multiplications (UMCM). The conventional UMCM, however, may incur considerable area/power overhead due to the unified topology constraint among different filter sets. To address this problem, we propose a new approximated MCM (AMCM) problem to relax the constraint and an approximate compatible graph synthesis (A-CGS) framework to efficiently solve AMCM by jointly searching for approximated filters and constructing a unified graph. Additionally, we suggest a lightweight finetuning method by freezing approximated filters and only finetuning biases, which can recover the original model’s accuracy within a few epochs. Experimental results with synthetic data demonstrate that AMCM reduces the area by up to 49.8%, 44.8%, and 40.3% when considering constant sets of 2, 4, and 8, respectively. Our designs with SR application achieve up to a 73.3% reduction in energy consumption. Experiments with the Set5 and Set14 datasets show that our model with bias correction achieves similar restoration performance compared to the eightbit models. © 1982-2012 IEEE.

关键词： Acceleration

来源：评论

学校读者我要写书评

暂无评论

vSpec: workload-adaptive operating system specialization for virtual machines in cloud computing

引用

Science China(Information Sciences) 2016年第9期59卷 47-62页

作者： Xinkui ZHAO Jianwei YIN Zuoning CHEN Sheng HE College of Computer Science Zhejiang University National Parallel Computing Engineering Research Center Jiangnan Institute of Computing Technology

In general, operating systems(OSs) are designed to mediate access to device hardware by applications. They process different kinds of system calls using an indiscriminate kernel with the same *** in cloud computing platforms are constructed from service components. Each of the service components is assigned separately to an individual virtual machine(VM), which leads to homogeneous system calls on each VM. In addition, the requirements for kernel function and configuration of system parameters from different VMs are different. Therefore, the suit-to-all design incurs an unnecessary performance overhead and restricts the OS’s processing capacity in cloud computing. In this paper, we propose an adaptive model for cloud computing to resolve the conflict between generality and performance. Our model adaptively specializes the OS of a VM according to the resource-consuming characteristics of workloads on the VM. We implement a prototype of the adaptive model, v Spec. There are five classes of VM: CPU-intensive, memory-intensive, I/O-intensive, networkintensive and compound, according to the resource-consuming characteristics of the workloads running on the VMs. v Spec specializes the OS of a VM according to the VM class. We perform comprehensive experiments to evaluate the effectiveness of v Spec on benchmarks and real-world applications.

关键词： workload classification operating system specialization workload-aware cloud computing v Spec system performance

来源：评论

学校读者我要写书评

暂无评论

Towards Cost-Optimal Policies for DAGs to Utilize IaaS Clouds with Online Learning

引用

IEEE Transactions on Services Computing 2025年

作者： Wu, Xiaohu Yu, Han Casale, Giuliano Gao, Guanyu Beijing University of Posts and Telecommunications National Engineering Research Center of Mobile Network Technologies Beijing China Nanyang Technological University WeBank-NTU Joint Research Centre on Fintech Singapore Singapore Nanyang Technological University School of Computer Science and Engineering Singapore Singapore Imperial College London Department of Computing London United Kingdom Nanjing University of Science and Technology School of Computer Science and Engineering Nanjing China

Premier cloud service providers (CSPs) offer on-demand and spot instances (i.e., virtual machines) with time-varying features in availability and price. While interacting with a CSP, what concerns users is the process of cost-effectively utilizing these instances, possibly in addition to self-owned instances. A job in data-intensive applications is represented by a directed acyclic graph (DAG) whose nodes represent tasks and the DAG can further be transformed into a chain of tasks. The key to achieving cost efficiency is determining the allocation of a specific deadline to each task, as well as the allocation of different types of instances to the task. In this paper, we make some mild assumptions on the usage of various instances and give an analytical understanding of the expected behaviors while utilizing instances. Based on this, we propose informed heuristic policies to determine the allocation of deadlines and instances. The policies are parametric to support the usage of online learning to infer the optimal values against the dynamics of cloud markets. Finally, intuitive greedy policies are used as baselines to validate the effectiveness of the proposed analytical solutions. The cost improvement is up to 23.58% when spot and on-demand instances are considered and up to 50.08% when self-owned instances are also considered. © 2008-2012 IEEE.

关键词： Cloud platforms

来源：评论

学校读者我要写书评

暂无评论

Exploring CLIP’s Dense Knowledge for Weakly Supervised Semantic Segmentation

arXiv

引用

arXiv 2025年

作者： Yang, Zhiwei Meng, Yucong Fu, Kexue Tang, Feilong Wang, Shuo Song, Zhijian Academy for Engineering and Technology Fudan University Shanghai200433 China Shanghai Key Laboratory of Medical Image Computing and Computer Assisted Intervention China Digital Medical Research Center School of Basic Medical Sciences Fudan University China Shandong Computer Science Center National Supercomputer Center in Jinan China

Weakly Supervised Semantic Segmentation (WSSS) with image-level labels aims to achieve pixel-level predictions using Class Activation Maps (CAMs). Recently, Contrastive Language-Image Pre-training (CLIP) has been introduced in WSSS. However, recent methods primarily focus on image-text alignment for CAM generation, while CLIP’s potential in patch-text alignment remains unexplored. In this work, we propose ExCEL to explore CLIP’s dense knowledge via a novel patch-text alignment paradigm for WSSS. Specifically, we propose Text Semantic Enrichment (TSE) and Visual Calibration (VC) modules to improve the dense alignment across both text and vision modalities. To make text embeddings semantically informative, our TSE module applies Large Language Models (LLMs) to build a dataset-wide knowledge base and enriches the text representations with an implicit attribute-hunting process. To mine fine-grained knowledge from visual features, our VC module first proposes Static Visual Calibration (SVC) to propagate fine-grained knowledge in a non-parametric manner. Then Learnable Visual Calibration (LVC) is further proposed to dynamically shift the frozen features towards distributions with diverse semantics. With these enhancements, ExCEL not only retains CLIP’s training-free advantages but also significantly outperforms other state-of-the-art methods with much less training cost on PASCAL VOC and MS COCO. Code is available at https://***/zwyang6/ExCEL. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

A New Median Filter Circuit Design Based on Atomic Silicon Quantum-Dot for Digital Image Processing and IoT Applications

引用

IEEE Internet of Things Journal 2025年

作者： Ahmadpour, Seyed-Sajad Avval, Danial Bakhshayeshi Navimipour, Nima Jafari Rasmi, Hadi Heidari, Arash Kassa, Sankit Misra, Neeraj Kumar Navin, Ahmad Habibizad Mosleh, Mohammad Hosseinzadeh, Mehdi Patidar, Mukesh Kadir Has University Department of Computer Engineering Istanbul Turkey Sakarya University Department of Information Systems Engineering Sakarya Turkey Department of Computer Engineering Kadir Has University Istanbul Turkey Future Technology Research Center National Yunlin University of Science and Technology Douliou Yunlin64002 Taiwan Western Caspian University Research Center of High Technologies and Innovative Engineering Baku Azerbaijan Islamic Azad University Department of Computer Engineering Dez.C. Dezful Iran İstanbul Atlas University Faculty of Engineering and Natural Science Department of Computer Engineering Istanbul Turkey Symbiosis International Deemed University Symbiois Institute of Technology Pune India VIT-AP University School of Electronics Engineering Andhra Pradesh Amaravathi522237 India Islamic Azad University Department of Computer Engineering Tabriz Branch Tabriz*** Iran Islamic Azad University Department of Computer Engineering Dezful Branch Dezful Iran Duy Tan University Institute of Research and Development Da Nang Viet Nam Indore Institute of Science and Technology Department of Electronics and Communication Engineering Madhya Pradesh Indore453111 India

Digital Image Processing (DIP) is the ability to manipulate digital photographs via algorithms for pattern detection, segmentation, enhancement, and noise reduction. In addition, the Internet of Things (IoT) acts as the eye and system for all DIP in various applications. It can possess a camera or another image sensor in order to capture real-time data from its environment. All vital data is processed by image processing in such a way that it recognizes the object, detects an anomaly, and automatically decides in real-time. In addition, in an IoT system, the median filter is the technique used for noise reduction by substituting the value of the pixel with the central value of the surrounding pixels. It provides speed and efficiency for quick analysis in all IoT systems. However, the images can get corrupted, especially in resource-constrained IoT devices with small cameras, because of random glitches. Moreover, using new quantum technology like atomic-scale silicon dangling bond (DB) logic circuits, which have advanced in fabrication and become a strong contender for field-coupled nano-computing, can solve previous problems in IoT systems. In this paper, we propose a unique quantum CSM based on two new proposed Mux and De-mux. The proposed CSM can be used for computational circuits like median filter circuits (MFC) in a wide range of digital circuits, specifically IoT devices. The proposed design is verified and validated using the powerful SiQAD tool. When comparing CSM to the newest designs, the suggested quantum circuit uses 85% less energy and takes up 61% less area. © 2014 IEEE.

关键词： Dangling bonds

来源：评论

学校读者我要写书评

暂无评论

Prognostication of zooplankton-driven cholera pathoepidemiological Dynamics: Novel Bayesian-regularized deep NARX neuroarchitecture

引用

computers in Biology and Medicine 2025年 192卷 110197-110197页

作者： Raja, Muhammad Junaid Ali Asif Sultan, Adil Chang, Chuan-Yu Shu, Chi-Min Kiani, Adiqa Kausar Shoaib, Muhammad Raja, Muhammad Asif Zahoor Department of Computer Science and Information Engineering National Yunlin University of Science and Technology 123 University Road Section 3 Douliou Yunlin64002 Taiwan Department of Safety Health and Environmental Engineering National Yunlin University of Science and Technology 123 University Road Section 3 Douliou Yunlin64002 Taiwan Future Technology Research Center National Yunlin University of Science and Technology 123 University Road Section 3 Douliou Yunlin64002 Taiwan AI Center Yuan Ze University Taoyuan320 Taiwan

Background: Cholera outbreaks pose significant health concerns, particularly through freshwater contamination through zooplankton serving as reservoirs for Vibrio Cholerae. Understanding these complex interactions within the aquatic ecosystem through mathematical modeling regimes may help us predict and prevent the spread of Cholera disease spread in affected regions. Method: In this study, an innovative Bayesian regularized deep nonlinear autoregressive exogenous (BRDNARX) neural networks are employed to model the intricate dynamics of Zooplankton-Driven Cholera Disease Transmission (ZDCDT) system. The cholera epidemic propagation through freshwater contamination is uncovered with analysis on densities of phytoplankton, vibrio cholerae carrying zooplankton, human population vector and microbial pathogen vector populous in the marine biosphere. Synthetic data for the ZDCDT is presented for diverse simulations using a modified Adams-Bashforth-Moulton predictor corrector numerical scheme. Subsequently, these temporal data sequences are preprocessed for the novel BRDNARX computing paradigm with an exhaustive assessment on mean square error iterative convergence plots, error histogram charts, regression index reports, input-error crosscorrelation charts, error autocorrelation charts, and time-series response dynamics. Results and conclusions: Comparative absolute error analysis with reference numerical solution adheres to diminutive disparities of range 10−3 to 10−9. Finally, BRDNARX neurostructures are reconfigured for predictive analysis of ZDCDT system in terms of single and multi-step ahead predictors with mean square error outcomes that range from 10−9 to 10−11. This establishes the efficacy of BRDNARX in correctly adhering to the intricacies of the zooplankton-driven cholera pathoepidemiological dynamics with precise forward prognostication. © 2025 Elsevier Ltd

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

CodecFake+: A Large-Scale Neural Audio Codec-Based Deepfake Speech Dataset

arXiv

引用

arXiv 2025年

作者： Chen, Xuanjun Du, Jiawei Wu, Haibin Zhang, Lin Lin, I. Ming Chiu, I. Hsiang Ren, Wenze Tseng, Yuan Tsao, Yu Jang, Jyh-Shing Roger Lee, Hung-Yi Graduate Institute of Communication Engineering National Taiwan University Taipei10617 Taiwan Department of Computer Science and Information Engineering National Taiwan University Taipei10617 Taiwan Brno University of Technology Brno61266 Czech Republic Department of Electrical Engineering National Taiwan University Taipei10617 Taiwan Research Center for Information Technology Innovation Academia Sinica Taipei11529 Taiwan

With the rapid advancement of neural audio codecs, codec-based speech generation (CoSG) systems have become highly powerful. Unfortunately, CoSG also enables the creation of highly realistic deepfake speech, making it easier to mimic an individual’s voice and spread misinformation. We refer to this emerging deepfake speech generated by CoSG systems as CodecFake. Detecting such CodecFake is an urgent challenge, yet most existing systems primarily focus on detecting fake speech generated by traditional speech synthesis models and fail to detect CodecFake speech. In this paper, we introduce CodecFake+, a large-scale dataset designed to advance codec-based deepfake speech detection and support the anti-spoofing community. To our knowledge, CodecFake+ is the largest dataset designed for this purpose, and encompasses the most diverse range of neural audio codec architectures. The training set is generated through re-synthesis using 31 publicly available open-source neural audio codec models, while the evaluation set includes web-sourced data from 17 advanced CoSG models. We also propose a novel and comprehensive neural audio codec taxonomy that categorizes codecs by their root components: vector quantizer, auxiliary objectives, and decoder types. Our proposed dataset and taxonomy enable detailed analysis at multiple levels to discern the key factors for successful CodecFake detection. At the individual codec level, we validate the effectiveness of using codec re-synthesized speech (CoRS) as training data for large-scale CodecFake detection. At the taxonomy level, we show that detection performance is strongest when the re-synthesis model incorporates disentanglement auxiliary objectives or a frequency-domain decoder. Furthermore, from the perspective of using all the CoRS training data, we show that our proposed taxonomy can be used to select better training data. By balancing training data with respect to decoder types, we improve detection performance and outperform mod

关键词： Taxonomies

来源：评论

学校读者我要写书评

暂无评论

Value Stream Mapping for Smart Pharmaceutical Management in a Thai Hospital

引用

Procedia computer Science 2025年 253卷 495-504页

作者： Ganda Boonsothonsatit Rardchawadee Silapunt Supachai Vongbunyong Kamol Kaemarungsi Wirachchaya Chanpuypetch Nilubon Chonsawat Graduate School of Management and Innovation King Mongkut’s University of technology Thonburi Thailand Electronic and Telecommunication Engineering King Mongkut’s University of Technology Thonburi Thailand Hospital Automation Research Center Institute of Field Robotics King Mongkut’s University of Technology Thonburi Thailand National Electronics and Computer Technology Center National Science and Technology Development Agency Thailand Faculty of Agro-Industry Chiang Mai University Thailand Department of Agricultural Economy and Developement Faculty of Agriculture Chiang Mai University Thailand

The evolution of healthcare systems worldwide necessitates continual improvement in hospital management practices, particularly pharmaceutical management. This paper explores transforming traditional pharmaceutical management systems in large and medium-sized hospitals in Thailand into smart systems using Value Stream Mapping (VSM) as a guiding tool. Traditional systems often face challenges such as inefficiencies and medication errors. VSM, rooted in lean management principles, offers a structured approach to visualize and optimize processes. Additionally, smart technologies like automated inventory systems and electronic health records enhance pharmaceutical management effectiveness. Through a comprehensive research methodology encompassing case studies, data collection, and VSM analysis, this study identifies inefficiencies and proposes evidence-based strategies for improvement. Results from a case study highlight the benefits of VSM implementation in enhancing pharmacy operations, emphasizing improved patient care and operational efficiency. The implementation of a smart medication dispensing system shows promising results in reducing workload and accommodating increased patient arrivals. Continued evaluation and refinement are crucial for maximizing benefits and addressing emerging challenges effectively.

关键词： lean management smart medication dispensing smart pharmaceutical smart technologies Value Stream Mapping

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Multi-Objective Spiking Neural Architecture Search for Image Classification

引用

IEEE Transactions on Evolutionary Computation 2025年

作者： Song, Xiaotian Lv, Zeqiong Fan, Jiaohao Deng, Xiong Lv, Jiancheng Liu, Jiyuan Sun, Yanan Sichuan University College of Computer Science Chengdu610065 China Stevens Institute of Technology Department of Mechanical Engineering HobokenNJ07030 United States Sichuan University State Key Laboratory of Oral Diseases National Center for Stomatology National Clinical Research Center for Oral Diseases Sichuan Provincial Engineering Research Center of Smart Diagnosis and Treatment for Oral Diseases Department of Oral Surgery West China Hospital of Stomatology Sichuan Chengdu610041 China

Spiking neural networks (SNNs) have the merit of energy efficiency, and have been widely used for various real-world applications. Similar to other types of neural networks, the performance of SNN is also significantly decided by its architecture. In this paper, we propose an Evolutionary Multi-objective Spiking Neural Architecture Search (EMO-SNAS) method that completely enables the automatic design of SNN architectures with both high performance and low power consumption. To achieve this, we first design a variable-length encoding strategy for SNNs, addressing the issue that traditional encoding strategies need to manually set the depth in advance. Furthermore, we propose an exploitation operator focusing on the local search for the variable-length encoding, as well as an exploration operator focusing on the global search based on the temporal expansion. Based on NSGA-II, EMO-SNAS can greatly balance the performance and power consumption during the architecture design. Experiments on three widely used image classification datasets show that EMO-SNAS can achieve the best among the state-of-the-art methods. Specifically, EMO-SNAS gains 0.45%, 0.26%, and 7.52% in terms of classification accuracy, yet significantly contributes to 33%, 29%, and 12% fewer spike numbers on CIFAR10, CIFAR100, and TinyImageNet datasets. Ablation studies show that the temporal expansion can improve the performance of EMO-SNAS. Moreover, the measurement of power consumption and theoretical convergence of EMO-SNAS are also discussed to justify its component design. In addition, with EMO-SNAS, the impact of initial channels for SNNs is also systemically investigated, based on which a conclusion against existing consensus is achieved. © 1997-2012 IEEE.

关键词： Encoding (symbols)

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：