检索结果-内蒙古大学图书馆

2025 IEEE international conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Li, Rui Yang, Tao Liu, Cheng Jiang, Dazhi Wong, Hau-San Wu, Si Department of Computer Science Shantou University China Department of Computer Science City University of Hong Kong Hong Kong School of Computer Science and Engineering South China University of Technology China

ISBN: (纸本)9798350368741

Pre-trained language models (PrLMs) demonstrate impressive performance on the sentiment analysis task. However, the large number of trainable parameters brings about heavy computational costs, which become more serious in multi-domain scenarios. In this paper, we propose to extract multi-layer features from the PrLM for efficient training since the training process is independent to its large backbone. Meanwhile, compared with the conventional feature extraction, we leverage prompts to induce PrLM for generating sentiment-aware features which lead to significant improvement on the sentiment analysis. In addition, most previous methods adopted a domain alignment paradigm for multi-domain learning, which becomes cumbersome when the number of domains is large. Therefore, we propose a novel prompt-augmented cross-domain contrastive learning for generalizable performance, which clusters samples with the same label under different prompts or domains. Our method is evaluated on two public multi-domain sentiment analysis benchmarks, which significantly outperforms recent state-of-the-art methods. Extensive ablation studies also verify the effectiveness of each proposed component. © 2025 IEEE.

关键词： multi-domain learning pre-trained language model prompt augmentation Sentiment analysis sentiment-aware feature

来源：评论

学校读者我要写书评

暂无评论

DigiEMine: Towards Leveraging Decision Mining and Context Data for Quality Control 28th

DigiEMine: Towards Leveraging Decision Mining and Context D...

引用

28th international conference on Enterprise Design, Operations, and Computing, EDOC 2024

作者： Wais, Beate Rinderle-Ma, Stefanie Faculty of Computer Science Research Group Workflow Systems and Technology University of Vienna Vienna Austria UniVie Doctoral School Computer Science DoCS University of Vienna Vienna Austria TUM School of Computation Information and Technology Technical University of Munich Garching Germany

ISBN: (纸本)9783031783371

Quality control processes in manufacturing often still rely on manual tasks. Applying decision mining can support users by providing valuable insight into the process. This paper discusses the potential of integrating contextual information into decision mining to achieve accurate and meaningful decision rules in the context of a case study stemming from the manufacturing domain. To explore this, a new approach, DigiEMine, is presented, which addresses the gap between information extraction and practical decision mining applications by integrating information extracted from engineering drawings with time sequence data in the form of diameter measurements of workpieces. The discovery of relational decision rules is enabled, allowing for contextualization of the decision rules. The output of this approach is presented in both textual decision rules and visually on engineering drawings, empowering users to make informed quality control decisions. The case study includes three datasets originating from cylindrical workpiece production. Results demonstrate the feasibility of the approach and the ability to generate meaningful decision rules across the tested datasets. Its potential applicability extends beyond the presented case study, with conceivable scenarios in multiple domains, such as healthcare or logistics, where integrating context information, such as regulatory data, with time sequence data is required to provide additional context for decisions. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Data quality

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-Based Traffic Information Prediction Methods in the Internet of Vehicles 14th

Deep Learning-Based Traffic Information Prediction Methods i...

引用

14th EAI international conference on Wireless and Satellite Systems, WiSATS 2024

作者： He, Chenguang Zhang, Bohan Ye, Liang Tan, Hua School of Electronics and Information Engineering Harbin Institute of Technology Harbin150001 China

ISBN: (纸本)9783031862021

In recent years, the rapid development of deep learning technology has provided many methods to predict traffic information prediction. In the Internet of Vehicles (IoV), accurate and real-time traffic information prediction plays an important role in improving system performance and user experience. How to effectively capture the temporal and spatial dependencies of traffic information is a major challenge in this field. In this paper, we focus on three neural network models (GRU, TGCN and TGCN-att) for traffic information prediction and train these three models using real datasets. We analyzed the outputs of each model separately, compared the performance metrics such as mean square error (RMSE) and mean absolute error (MAE) between the predicted and real values, and calculated the accuracy of the predictions of each model. The simulation results show that since road networks generally have a complex topology, correctly capturing the spatial dependence between data is very important for improving the prediction accuracy of the models when performing traffic information prediction. © ICST Institute for computer sciences, Social Informatics and Telecommunications Engineering 2025.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

IoT-enabled Student Attendance System using RISC-V based Embedded System 22

IoT-enabled Student Attendance System using RISC-V based Emb...

引用

22nd international Learning and technology conference, L and T 2025

作者： Khan, Khadija Alzahrani, Nada Abed, Reham Razi, Iman Sofia Dhaou, Imed Ben Dar Al-Hekma University Hekma School of Engineering Computing and Design Department of Computer Science Jeddah22246-4872 Saudi Arabia

ISBN: (纸本)9798331532215

Inefficiencies, cumbersome processes, and a significant risk of errors, delays, and misuse often plague conventional methods for tracking student attendance. To address these challenges, this paper introduces a smart monitoring system driven by the Artificial Intelligence of Things (AIoT), specifically designed for resource-limited embedded processors. This research examines a real-time facial recognition-based student attendance tracking system implemented on an edge device. Additionally, it assesses the effectiveness of ARM and RISC-V microprocessors as the foundation for a reliable edge computing platform. The system employs Python's Dlib library and the ResNet model for effective facial recognition. Comparisons between the Raspberry Pi Model 4 B and the RISC-V StarFive VisionFive2 are made to determine the optimal platform for extensive deployments. The proposed edge system emphasizes security, automation, and reliability, facilitating real-time, precise attendance tracking and significantly improving operational efficiency for educators and administrators. The solution using the RISC-V processor is at least 10 times more energy-efficient than the competing solution based on the RISC processor instruction set architecture. © 2025 IEEE.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

Analyzing End-of-Course Comments by Using Deep Hierachical Net

Analyzing End-of-Course Comments by Using Deep Hierachical N...

引用

Electronics, Energy Systems and Power Engineering (EESPE), IEEE international conference on

作者： Fei Hu School of Computer Science and Engineering (School of Artificial Intelligence) Chongqing University of Science and Technology Chongqing China

ISBN: (数字)9798350389579

ISBN: (纸本)9798350389586

The extensive feedback that students left at the end of courses can be very helpful in enhancing and honing the instruction. The comments, expressed in text format, cannot be effectively represented in a way that the computer can recognize, and the meaning implied in the comments is always blurry, making it more difficult for the computer to understand them. As a result, it is challenging for the computer to use these comments and comprehend student appeal. In this paper, we address the representation problem by projecting comments (including words) into a dense vector space. We suggest using a highly deep neural network to capture the textual meaning while examining a real-world dataset of five-year comments from end-of-course evaluations at a university. The suggested model is assessed based on its capacity to identify the sentimental inclinations of these remarks, yielding insightful recommendations and critiques.

关键词： Power engineering Text recognition Artificial neural networks Aerospace electronics Vectors

来源：评论

学校读者我要写书评

暂无评论

HARTS: High-Threshold, Adaptively Secure, and Robust Threshold Schnorr Signatures 30th

HARTS: High-Threshold, Adaptively Secure, and Robust Thresho...

引用

30th international conference on the Theory and Application of Cryptology and Information Security

作者： Bacho, Renas Loss, Julian Stern, Gilad Wagner, Benedikt CISPA Helmholtz Ctr Informat Secur Saarbrucken Germany Tel Aviv Univ Tel Aviv Israel Saarland Univ Saarbrucken Germany Ethereum Fdn Berlin Germany

ISBN: (纸本)9789819608904;9789819608911

Threshold variants of the Schnorr signature scheme have recently been at the center of attention due to their applications to cryptocurrencies. However, existing constructions for threshold Schnorr signatures among a set of n parties with corruption threshold tc suffer from at least one of the following drawbacks: (i) security only against static (i.e., non-adaptive) adversaries, (ii) cubic or higher communication cost to generate a single signature, (iii) strong synchrony assumptions on the network, or (iv) t(c) + 1 are sufficient to generate a signature, i.e., the corruption threshold of the scheme equals its reconstruction threshold. Especially (iv) turns out to be a severe limitation for many asynchronous realworld applications where t(c) < n/3 is necessary to maintain liveness, but a higher signing threshold of n - t(c) is needed. A recent scheme, ROAST, proposed by Ruffing et al. (ACM CCS '22) addresses (iii) and (iv), but still falls short of obtaining subcubic complexity and adaptive security. In this work, we present HARTS, the first threshold Schnorr signature scheme to incorporate all these desiderata. More concretely: - HARTS is adaptively secure and remains fully secure and operational even under asynchronous network conditions in the presence of up to t(c) < n/3 malicious parties. This is optimal. - HARTS outputs a Schnorr signature of size lambda with a near-optimal amortized communication cost of O(lambda n(2) log n) bits and a single online round per signature. - HARTS is a high-threshold scheme: no fewer than t(r) + 1 signature shares can be combined to yield a full signature, where any t(r) is an element of[t(c), n-t(c)) is supported. This especially covers the case t(r) >= 2n/3 > 2t(c). This is optimal. We prove our result in a modular fashion in the algebraic group model. At the core of our construction, we design a new simple and adaptively secure high-threshold asynchronous verifiable secret sharing (AVSS) scheme which may be of independent in

关键词： Threshold Signatures Schnorr Signatures Adaptive Security Robustness High-Threshold Asynchronous network

来源：评论

学校读者我要写书评

暂无评论

Draw Sketch, Draw Flesh: Whole-Body Computed Tomography from Any X-Ray Views

引用

international JOURNAL OF computer VISION 2025年第5期133卷 2505-2526页

作者： Pan, Yongsheng Ye, Yiwen Zhang, Yanning Xia, Yong Shen, Dinggang Polytech Univ Sch Comp Sci & Engn Natl Engn Lab Integrated Aerosp Ground Ocean Big D Xian 710129 Shaanxi Peoples R China Northwestern Polytech Univ Shenzhen Res Inst Shenzhen 518057 Guangdong Peoples R China Northwestern Polytech Univ Ningbo Res Inst Ningbo 315000 Zhejiang Peoples R China ShanghaiTech Univ Sch Biomed Engn Shanghai 201210 Peoples R China Shanghai United Imaging Intelligence Co Ltd Shanghai 200232 Peoples R China Shanghai Clin Res & Trial Ctr Shanghai 201210 Peoples R China

Stereoscopic observation is a common foundation of medical image analysis and is generally achieved by 3D medical imaging based on settled scanners, such as CT and MRI, that are not as convenient as X-ray machines in some flexible scenarios. However, X-ray images can only provide perspective 2D observation and lack view in the third dimension. If 3D information can be deduced from X-ray images, it would broaden the application of X-ray machines. Focus on the above objective, this paper dedicates to the generation of pseudo 3D CT scans from non-parallel 2D perspective X-ray (PXR) views and proposes the Draw Sketch and Draw Flesh (DSDF) framework to first roughly predict the tissue distribution (Sketch) from PXR views and then render the tissue details (Flesh) from the tissue distribution and PXR views. Different from previous studies that focus only on partial locations, e.g., chest or neck, this study theoretically investigates the feasibility of head-to-leg reconstruction, i.e., generally applicable to any body parts. Experiments on 559 whole-body samples from 4 cohorts suggest that our DSDF can reconstruct more reasonable pseudo CT images than state-of-the-art methods and achieve promising results in both visualization and various downstream tasks. The source code and well-trained models are available a https://***/YongshengPan/WholeBodyXraytoCT.

关键词： Image synthesis Generative adversarial network X-ray images CT reconstruction

来源：评论

学校读者我要写书评

暂无评论

Improved Stroke Lesion Segmentation via Cross-Model Knowledge Distillation 4th

Improved Stroke Lesion Segmentation via Cross-Model Knowled...

引用

4th international Workshop on Imaging and Treatment Challenges, SWITCH 2024, and 6th international Challenge on Ischemic Stroke Lesion Segmentation Challenge, ISLES 2024, Held in Conjunction with Medical Image Computing and computer Assisted Intervention, MICCAI 2024

作者： Liu, Zixin Pang, Haowen Zhang, Xinru Ye, Chuyang School of Integrated Circuits and Electronics Beijing Institute of Technology Beijing China

ISBN: (纸本)9783031811005

Stroke lesion segmentation is a crucial step in stroke disease analysis as it provides essential anatomical information for the diagnosis and prognosis. Deep learning approaches, including the convolutional neural network (CNN) and Transformer, have improved the accuracy of automated stroke lesion segmentation. Since CNN-based models generally learn to extract local features with convolutional filters, whereas Transformer-based models excel in modeling long-range dependencies with self-attention mechanisms, combining the strengths of the two types of models may bring additional benefits. Existing studies have proposed various hybrid network structures combining convolution and self-attention;however, empirical evidence shows that these hybrid structures do not necessarily outperform purely CNN-based or Transformer-based methods. Thus, this work further explores the integration of the strengths of CNNs and Transformers. Instead of constructing a hybrid network architecture, we propose a cross-model interaction method based on knowledge distillation, which can effectively allow CNN-based and Transformer-based models to learn the strength of each other. The guidance is achieved with an adaptive recall-enhancing loss for knowledge distillation, which suppresses negative knowledge transfer and encourages the model to reduce false negative predictions that are common in stroke lesion segmentation. We validated the proposed method on two public stroke lesion segmentation datasets, where it improves the segmentation performance compared with the best competing segmentation model by 2.0% and 1.4% for the ISLES and ATLAS dataset, respectively. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Research and Application of Customer Side Security Energy Use Monitoring technology Based on Artificial Intelligence and Digital Power Room 7th

Research and Application of Customer Side Security Energy Us...

引用

7th international conference on Communication, Device and networking, ICCDN 2024

作者： Li, Jincan Dai, Ying Zhu, Wanting Li, Pei Yang, Xiaqin Liu, Ying Marketing Department of Guangxi Power Grid Co. Ltd. Guangxi Nanning530023 China

ISBN: (纸本)9789819764648

With the development of the power industry, smart grid has become the development direction of the future global power grid. With the help of a new generation of information technology, the collection of power consumption data at the power terminal, and the analysis of power consumption data, and then timely prediction or identification of electrical faults, timely warning to the user or automatic cut off the power supply, can effectively reduce the risk of electrical faults caused by accidents, reduce the loss caused by accidents. Therefore, this paper mainly introduces the development and implementation of intelligent electricity safety monitoring system and fault arc identification algorithm. Firstly, the intelligent electricity safety monitoring system is designed. The main functions of the system are as follows: terminal data collection function, LoRa network wireless transmission function, intelligent gateway data exchange function, server human–computer interaction function and intelligent processing algorithm function. Finally, through the system test, it is found that the intelligent electricity monitoring system can identify the fault arc with 96% accuracy, which is suitable for customer-side safety energy monitoring. Through the research and application of intelligent electricity safety monitoring technology, users can grasp the status information and operation of electrical equipment in real time, and establish a perfect operation and maintenance management system based on intelligent electricity monitoring technology. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Gateways (computer networks)

来源：评论

学校读者我要写书评

暂无评论

Toward the Implementation of a Cooking Support System Complementing Nonexistent Objects with Virtual Objects 22nd

Toward the Implementation of a Cooking Support System Com...

引用

22nd international conference on Advances in Mobile Computing and Multimedia Intelligence, MoMM 2024

作者： Nihanda, Taiki Sano, Shoji Graduate School of Engineering Kanazawa Institute of Technology Ishikawa Nonoichi Japan College of Engineering Kanazawa Institute of Technology Ishikawa Nonoichi Japan

ISBN: (纸本)9783031780486

Cooking requires the usage of various ingredients and cooking utensils. However, if we lack some of them due to insufficient preparation, we cannot cook as expected. In this paper, we propose a cooking support system that complements nonexistent objects with virtual objects. We developed a prototype for cutting apples to evaluate the differences between using the real object and the virtual one. The results showed that the cutting technique equally enhanced both the cases of using a real knife and a virtual knife. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Virtual environments

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：