检索结果-内蒙古大学图书馆

14th International Conference on Information and Communication technology Convergence, ICTC 2023

作者： Park, Jiwoo Jwa, Younkyung Kwak, Jiin Lim, Jihun Kim, Sehee Pohang University of Science and Technology Ai Graduate School Pohang Korea Republic of Gwangju Institute of Science and Technology Ai Graduate School Gwangju Korea Republic of Ulsan National Institute of Science and Technology Ai Graduate School Ulsan Korea Republic of Korea University Department of Mathematics Seoul Korea Republic of Seoul National University Department of Computer Science and Engineering Seoul Korea Republic of

ISBN: (纸本)9798350313277

Due to soccer's immense popularity, numerous soccer matches are played each year and people watch or record these matches. However, considering that the duration of a game of soccer is at least 90 minutes and that important scenes such as goals and passes take up a small portion of the total time, we inevitably look for highlights rather than the entire game. Therefore, we propose a model that automatically generates a highlight video of a full soccer match in this paper. For this purpose, we searched diverse action-spotting models to choose a backbone model for our highlight classifier. Through various experiments, NetVLAD++ is selected as a backbone model and we created a fairly natural highlight generator by combining NetVLAD++ and our highlight classifier. Our results are meaningful because they can be applied in various fields and have huge room for improvement. © 2023 IEEE.

关键词： Automatic Video Summarization Highlight Generation Soccer Video Highlight

来源：评论

学校读者我要写书评

暂无评论

Learn from Balance: Rectifying Knowledge Transfer for Long-Tailed Scenarios

Learn from Balance: Rectifying Knowledge Transfer for Long-T...

引用

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

作者： Xinlei Huang Jialiang Tang Xubin Zheng Jinjia Zhou Wenxin Yu Ning Jiang School of Computer Science and Technology Southwest University of Science and Technology School of Information Science and Technology Great Bay University School of Computer Science and Engineering Nanjing University of Science and Technology Graduate School of Science and Engineering Hosei University

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Knowledge Distillation (KD) transfers knowledge from a large pre-trained teacher network to a compact and efficient student network, making it suitable for deployment on resource-limited media terminals. However, traditional KD methods require balanced data to ensure robust training, which is often unavailable in practical applications. In such scenarios, a few head categories occupy a substantial proportion of examples. This imbalance biases the trained teacher network towards the head categories, resulting in severe performance degradation on the less represented tail categories for both the teacher and student networks. In this paper, we propose a novel framework called Knowledge Rectification Distillation (KRDistill) to address the imbalanced knowledge inherited in the teacher network through the incorporation of the balanced category priors. Furthermore, we rectify the biased predictions produced by the teacher network, particularly focusing on the tail categories. Consequently, the teacher network can provide balanced and accurate knowledge to train a reliable student network. Intensive experiments conducted on various long-tailed datasets demonstrate that our KRDistill can effectively train reliable student networks in realistic scenarios of data imbalance.

关键词： Knowledge engineering Training Heavily-tailed distribution Focusing Tail Signal processing Media Reliability Speech processing Knowledge transfer

来源：评论

学校读者我要写书评

暂无评论

Heart Disease Prediction Using Novel Ensemble and Blending Based Cardiovascular Disease Detection Networks: EnsCVDD-Net and BlCVDD-Net

引用

IEEE Access 2024年 12卷 109230-109254页

作者： Khan, Hira Javaid, Nadeem Bashir, Tariq Akbar, Mariam Alrajeh, Nabil Aslam, Sheraz Department of Electrical and Computer Engineering COMSATS University Islamabad Islamabad44000 Pakistan Department of Computer Science COMSATS University Islamabad Islamabad44000 Pakistan International Graduate School of Artificial Intelligence National Yunlin University of Science and Technology Yunlin Douliu64002 Taiwan Department of Biomedical Technology College of Applied Medical Sciences King Saud University Riyadh11633 Saudi Arabia Cyprus University of Technology Limassol3036 Cyprus

Cardiovascular Diseases (CVDs) have emerged as a significant physiological condition, being a primary contributor to mortality. Timely and precise diagnosis of heart disease is crucial to safeguard patients from additional harm. Recent studies show that the usage of data driven approaches, such as Deep Learning (DL) and Machine Learning (ML) techniques, in the field of medical science is highly useful in accurately diagnosing heart disease in less time. However, statistical learning and traditional ML approaches require feature engineering to generate robust and effective features from data, which are then used in the prediction models. In the case of large complex data, both processes pose many challenges. Whereas, DL techniques are capable of learning features automatically from the data and are effective at handling large and intricate datasets while outperforming the ML models. This study focuses on the accurate prediction of CVDs, considering the patient’s health and socio-economic conditions while mitigating the challenges presented by imbalanced data. The Adaptive Synthetic Sampling Technique is used for data balancing, while the Point Biserial Correlation Coefficient is used as a feature selection technique. In this study, two DL models, Ensemble based Cardiovascular Disease Detection Network (EnsCVDD-Net) and Blending based Cardiovascular Disease Detection Network (BlCVDD-Net), are proposed for accurate prediction and classification of CVDs. EnsCVDD-Net is made by applying an ensemble technique to LeNet and Gated Recurrent Unit (GRU), and BlCVDD-Net is made by blending LeNet, GRU and Multilayer Perceptron. SHapley Additive exPlanations is used to provide a clear understanding of the influence different factors have on CVD diagnosis. The network’s performance is evaluated on the basis of various performance metrics. The results indicate that the EnsCVDD-Net outperforms all base models with 88% accuracy, 88% F1-score, 91% precision, 85% recall, and 777s execu

关键词： Heart

来源：评论

学校读者我要写书评

暂无评论

STRUCTURAL RETROFITTING OF EXISTING REINFORCED CONCRETE GIRDERS BY BEAM STRINGS

引用

AIJ Journal of technology and Design 2022年第69期28卷 655-660页

作者： Shimizu, Keiichi Kato, Daisuke Nakamura, Takaya Graduate School of Science and Technology Niigata Univ. Japan Architecture Program Faculty of Eng. Niigata Univ. Japan

This report shows the structural retrofitting of a two-story reinforcedconcrete (RC) welfare-facility building in Niigata prefecture, which was built in 2002. During the Chuetsu earthquake in 2004, the first floor window under the bathroom was damaged. Our investigation revealed that the damage was caused by an error in the structural design and insufficient girder strengths. Thus, such RC girders were retrofitted using the beam string structures composing steel members. The stress was reduced by introducing a pushing force to satisfy the allowable stress. This method was simple, easy to install, and economical. © 2022 Architectural Institute of Japan. All rights reserved.

关键词： Reinforced concrete

来源：评论

学校读者我要写书评

暂无评论

MARS: An Adaptive Multi-Agent DRL-based Scheduler for Multipath QUIC in Dynamic Networks 31

MARS: An Adaptive Multi-Agent DRL-based Scheduler for Multip...

引用

31st IEEE/ACM International Symposium on Quality of Service, IWQoS 2023

作者： Han, Xueqiang Han, Biao Li, Ruidong Ji, Xiaolan College of Computer National University of Defense Technology Changsha China Graduate School of Natural Science and Technology Kanazawa University Kanazawa Japan

ISBN: (纸本)9798350399738

The multipath extension of the Quick UDP Internet Connection (QUIC) protocol, also called MPQUIC, is currently attracting increasing attention from both industry and academia. The multipath scheduler of MPQUIC determines how to distribute the packets onto different paths. However, our experimental results show that they fail to adapt to various receive buffer sizes and Quality of Service (QoS) requirements while applying current multipath schedulers into MPQUIC due to the diversity of devices and applications. These problems are especially severe under heterogeneous and dynamic network environments. To tackle these problems, we propose MARS, a Multi-Agent deep Reinforcement learning (MADRL) based Multipath QUIC Scheduler, which is able to promptly adapt to dynamic network environments. It exploits the MADRL method to learn a neural network for each path and generate scheduling policy. Besides, it introduces a novel multi-objective reward function that takes out-of-order (OFO) queue size and different QoS metrics into consideration to realize adaptive scheduling optimization. We implement MARS in an MPQUIC prototype and compare it with the state-of-the-art multipath schedulers in both emulated and real-world networks. Experimental results show that MARS outperforms the other schedulers with better adaptive capability regarding the receive buffer sizes and QoS. © 2023 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Toward Digital Transformation Adoption: A Conceptual Framework from Transformational Leadership Perspective 7

Toward Digital Transformation Adoption: A Conceptual Framewo...

引用

7th Information Systems International Conference, ISICO 2023

作者： Marcel Gaol, Ford Lumban Supangkat, Suhono Harso Ranti, Benny Computer Science Department BINUS Graduate Program - Doctor of Computer Science Bina Nusantara University Jakarta Indonesia Smart Cities and Communities Innovation Center Faculty of Digital Communication and Hotel & Tourism Institute of Technology Bandung Bandung Indonesia Computer Science Department Faculty of Computer Science University of Indonesia Depok Indonesia Information System Department Universitas Kristen Krida Wacana Jakarta Indonesia

Digital transformation is about transforming processes, business models, domains, and culture. Studies show that the failure rate of digital transformation is quite high up to 90%. Studies show that the transformational leadership model has a significant impact on digital transformation adoption. This paper identifies the positive and negative attributes of transformational leadership including the components that support and are affected for successful adoption of digital transformation. Furthermore, the paper combines several findings related to the attributes and components in the form of a conceptual framework. The conceptual framework can serve as a guide for organizations for their digital transformation journey. © 2023 The Authors. Published by Elsevier B.V.

关键词： Failure rate

来源：评论

学校读者我要写书评

暂无评论

Analysis of Several Quantum Encoding Methods Implemented on A Quantum Circuit Architecture to Improve Classification Accuracy 6

Analysis of Several Quantum Encoding Methods Implemented on ...

引用

6th International Conference on Electrical, Telecommunication and computer Engineering, ELTICOM 2022

作者： Fauzi, Rahmad Zarlis, Muhammad Mawengkang, Herman Sihombing, Poltak Graduate Program of Computer Science Faculty of Computer Science and Information Technology Universitas Sumatera Utara Medan Indonesia Faculty of Computer Science and Information Technology Universitas Sumatera Utara Medan Indonesia

ISBN: (纸本)9798350320718

Quantum encoding is a process to transform classical information into quantum states. It plays a crucial role in using quantum algorithms to solve classical problems, especially in quantum machine learning tasks. There are many QE methods. It is very difficult to determine which QE method to choose to improve classification accuracy. Therefore, this paper will analyze several QE methods. Training and testing on Iris flower datasets were performed in a architecture quantum circuit and some performances parameters were evaluated. The expected result is that we can compare the classification accuracy of some of these Quantum encodings. © 2022 IEEE.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

MamMIL: Multiple Instance Learning for Whole Slide Images with State Space Models

MamMIL: Multiple Instance Learning for Whole Slide Images wi...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Fang, Zijie Wang, Yifeng Zhang, Ye Wang, Zhi Zhang, Jian Ji, Xiangyang Zhang, Yongbing Tsinghua University Tsinghua Shenzhen International Graduate School Shenzhen China Shenzhen China School of Computer Science and Technology Shenzhen China Peking University School of Electronic and Computer Engineering Shenzhen Graduate School Shenzhen China Tsinghua University Department of Automation Beijing China

ISBN: (纸本)9798350386226

Recently, pathological diagnosis has achieved superior performance by combining deep learning models with the multiple instance learning (MIL) framework using whole slide images (WSIs). However, the giga-pixeled nature of WSIs poses a great challenge for efficient MIL. Existing studies either do not consider global dependencies among instances, or use approximations such as linear attentions to model the pair-to-pair instance interactions, which inevitably brings performance bottlenecks. To tackle this challenge, we propose a framework named MamMIL for WSI analysis by cooperating the selective structured state space model (i.e., Mamba) with MIL, enabling the modeling of global instance dependencies while maintaining linear complexity. Specifically, considering the irregularity of the tissue regions in WSIs, we represent each WSI as an undirected graph. To address the problem that Mamba can only process 1D sequences, we further propose a topology-aware scanning mechanism to serialize the WSI graphs while preserving the topological relationships among the instances. Finally, in order to further perceive the topological structures among the instances and incorporate short-range feature interactions, we propose an instance aggregation block based on graph neural networks. Experiments show that MamMIL can achieve advanced performance than the state-of-the-art frameworks. The code can be accessed at https://***/Vison307/MamMIL. © 2024 IEEE.

关键词： Undirected graphs

来源：评论

学校读者我要写书评

暂无评论

Interactive Training System for Electric Guitar Strumming Form by Using Inertial Motion Sensors 11

Interactive Training System for Electric Guitar Strumming Fo...

引用

11th IEEE Global Conference on Consumer Electronics, GCCE 2022

作者： Matsushita, Soichiro Kamo, Fumiyoshi Tokyo University of Technology School of Computer Science Tokyo Hachioji Japan Tokyo University of Technology Graduate School of Computer Science Tokyo Hachioji Japan

ISBN: (纸本)9781665492324

A 3-D CG-based graphical interaction system for electric guitar-playing lessons has been investigated by using a wrist-worn motion-sensing device. Assuming a virtual camera always looks straight to the guitar player's strumming hand, stable and intuitive 3-D CG graphics expressing the guitar playing form can be obtained onto a laptop PC. The camera angle from the head of the guitar towards the strumming hand may give suggestive information on the correct playing form taught by a professional guitarist. A 14-week-long population test with 12 amateur players showed that a large elbow-swinging angle amplitude is crucial to achieve sharp chord strumming. A strumming form evaluation utilizing a template matching calculation with the elbow-swinging and the wrist-twisting angles showed how the skill development had been done. © 2022 IEEE.

关键词： Template matching

来源：评论

学校读者我要写书评

暂无评论

Evaluating Strength of Piano Key Touch by Using a Wrist-Worn Inertial Motion Sensor 11

Evaluating Strength of Piano Key Touch by Using a Wrist-Worn...

引用

11th IEEE Global Conference on Consumer Electronics, GCCE 2022

作者： Komatsu, Kaname Kamo, Fumiyoshi Matsushita, Soichiro Tokyo University of Technology School of Computer Science Tokyo Hachioji Japan Tokyo University of Technology Graduate School of Computer Science Tokyo Hachioji Japan

ISBN: (纸本)9781665492324

The strength of the piano key touch plays important role in musical expression. Although it is possible to assess the key touch strength by hearing the sound, it would be still unclear how it is achieved by the playing motion. A wrist-worn device utilizing a commercially available inertial motion sensor has been investigated to evaluate the strength of key touch without using cameras ether additional wearable devices. As a result, a strong correlation coefficient of R = 0.95 was found between values of the MIDI velocity from an electronic piano while playing a simple C-chord and the peak angular velocity towards the key pressing direction with 9 subjects having a wide range of piano playing experience. In addition, the angular velocity-based estimation was found to be effective in the weak key touch region. A time-integral value of the linear acceleration toward the ground showed superior characteristics in terms of estimating the MIDI velocity in the strong key touch region. © 2022 IEEE.

关键词： Angular velocity

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：