检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Zhou, Qingyuan Yang, Weidong Fei, Ben Xu, Jingyi Zhang, Rui Liu, Keyi Luo, Yeqi He, Ying School of Computer Science Fudan University China Department of Information Engineering The Chinese University of Hong Kong Hong Kong College of Computing and Data Science Nanyang Technological University Singapore

Noise is an inevitable aspect of point cloud acquisition, necessitating filtering as a fundamental task within the realm of 3D vision. Existing learning-based filtering methods have shown promising capabilities on commonly used datasets. Nonetheless, the effectiveness of these methods is constrained when dealing with a substantial quantity of point clouds. This limitation primarily stems from their limited denoising capabilities for dense and large-scale point clouds and their inclination to generate noisy outliers after denoising. To deal with this challenge, we introduce 3DMambaIPF, for the first time, exploiting Selective State Space Models (SSMs) architecture to handle highly-dense and large-scale point clouds, capitalizing on its strengths in selective input processing and large context modeling capabilities. Additionally, we present a robust and fast differentiable rendering loss to constrain the noisy points around the surface. In contrast to previous methodologies, this differentiable rendering loss enhances the visual realism of denoised geometric structures and aligns point cloud boundaries more closely with those observed in real-world objects. Extensive evaluations on commonly used datasets (typically with up to 50K points) demonstrate that 3DMambaIPF achieves state-of-the-art results. Moreover, we showcase the superior scalability and efficiency of 3DMam-baIPF on highly dense and large-scale point clouds with up to 500K points compared to off-the-shelf methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Wiener filtering

来源：评论

学校读者我要写书评

暂无评论

Non-convergence to global minimizers in data driven supervised deep learning: Adam and stochastic gradient descent optimization provably fail to converge to global minimizers in the training of deep neural networks with ReLU activation

arXiv

引用

arXiv 2024年

作者： Hannibal, Sonja Jentzen, Arnulf Thang, Do Minh Applied Mathematics: Institute for Analysis and Numerics Faculty of Mathematics and Computer Science University of Münster Germany School of Data Science Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen [CUHK-Shenzhen China School of Data Science The Chinese University of Hong Kong Shenzhen [CUHK-Shenzhen China Department of Probability and Statistic Institute of Mathematics Vietnam Academy of Science and Technology Viet Nam

Deep learning (DL) methods – consisting of a class of deep neural networks (DNNs) trained by a stochastic gradient descent (SGD) optimization method – are nowadays key tools to solve data driven supervised learning problems. Despite the great success of SGD methods in the training of DNNs, it remains a fundamental open problem of research to explain the success and the limitations of such methods in rigorous theoretical terms. In particular, even in the standard setup of data driven supervised learning problems, it remained an open research problem to prove (or disprove) that SGD methods converge in the training of DNNs with the popular rectified linear unit (ReLU) activation function with high probability to global minimizers in the optimization landscape. In this work we answer this question negatively by proving that it does not hold that SGD methods converge with high probability to global minimizers of the objective function. Even stronger, in this work we prove for a large class of SGD methods that the considered optimizer does with high probability not converge to global minimizers of the optimization problem. It turns out that the probability to not converge to a global minimizer converges at least exponentially quickly to one as the width of the first hidden layer of the ANN (the number of neurons on the first hidden layer) and the depth of the ANN (the number of hidden layers), respectively, increase to infinity. The general non-convergence results of this work do not only apply to the plain vanilla standard SGD method but also to a large class of accelerated and adaptive SGD methods such as the momentum SGD, the Nesterov accelerated SGD, the Adagrad, the RMSProp, the Adam, the Adamax, the AMSGrad, and the Nadam optimizers. However, we would like to emphasize that the findings of this work do not imply that SGD methods do not succeed to train DNNs: it may still very well be the case that SGD methods provably succeed to train DNNs in data driven learning pr

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

On the origins of linear representations in large language models 24

On the origins of linear representations in large language m...

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Yibo Jiang Goutham Rajendran Pradeep Ravikumar Bryon Aragam Victor Veitch Department of Computer Science University of Chicago Machine Learning Department Carnegie Mellon University Booth School of Business University of Chicago Department of Statistics and Data Science Institute University of Chicago

Recent works have argued that high-level semantic concepts are encoded "linearly" in the representation space of large language models. In this work, we study the origins of such linear representations. To that end, we introduce a simple latent variable model to abstract and formalize the concept dynamics of the next token prediction. We use this formalism to show that the next token prediction objective (softmax with cross-entropy) and the implicit bias of gradient descent together promote the linear representation of concepts. Experiments show that linear representations emerge when learning from data matching the latent variable model, confirming that this simple structure already suffices to yield linear representations. We additionally confirm some predictions of the theory using the LLaMA-2 large language model, giving evidence that the simplified model yields generalizable insights.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multi-Grained Feature Pruning for Video-Based Human Pose Estimation

Multi-Grained Feature Pruning for Video-Based Human Pose Est...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Wang, Zhigang Fan, Shaojing Liu, Zhenguang Wu, Zheqi Wu, Sifan Jiao, Yingying College of Computer Science and Technology Zhejiang Gongshang University Hangzhou China School of Computing National University of Singapore Singapore The State Key Laboratory of Blockchain and Data Security Zhejiang University Hangzhou China Institute of Blockchain and Data Security Hangzhou China College of Computer Science and Technology Jilin University Changchun China College of Computer Science and Technology Zhejiang University of Technology Hangzhou China

Human pose estimation, with its broad applications in action recognition and motion capture, has experienced significant advancements. However, current Transformer-based methods for video pose estimation often face challenges in managing redundant temporal information and achieving fine-grained perception because they only focus on processing low-resolution features. To address these challenges, we propose a novel multi-scale resolution framework that encodes spatio-temporal representations at varying granularities and executes fine-grained perception compensation. Furthermore, we employ a density peaks clustering method to dynamically identify and prioritize tokens that offer important semantic information. This strategy effectively prunes redundant feature tokens, especially those arising from multi-frame features, thereby optimizing computational efficiency without sacrificing semantic richness. Empirically, it sets new benchmarks for both performance and efficiency on three large-scale datasets. Our method achieves a 93.8% improvement in inference speed compared to the baseline, while also enhancing pose estimation accuracy, reaching 87.4 mAP on the PoseTrack2017 dataset. © 2025 IEEE.

关键词： Large datasets

来源：评论

学校读者我要写书评

暂无评论

Leveraging Support Vector Machines for Detection of Network Traffic Attacks

Leveraging Support Vector Machines for Detection of Network ...

引用

Advancement in Electronics & Communication Engineering (AECE), International Conference on

作者： Anurag Tiwari School of Artificial Intelligence & Data Science Indian Institute of Technology Jodhpur India Department of Computer Science & Engineering Babu Banarasi Das Institute of Technology& Management Lucknow India

ISBN: (数字)9798350364729

ISBN: (纸本)9798350364736

Network security is crucial to ensure the protection of information systems in an increasingly connected world, and network traffic attack detection is an important aspect of maintaining system integrity. This study addresses this by investigating the use of a powerful and flexible machine learning algorithm, namely Support Vector Machines (SVM), to detect network traffic attacks. As a result, SVM has been successfully applied in different domains like network security, etc. In this study, SVM is applied for normal and malicious network traffic classification, including several attack types like Denial-of-Service (DoS), User-to-Root (U2R), and Remote-to-Local (R2L) attacks. To this end, the KDD Cup 99 dataset is employed for training and testing the model. These metrics are then used to evaluate the performance of the SVM-based intrusion detection system. The results demonstrate that SVM is able to recognize a high level of accurateness on network data traffic attacks as well as it has performed well in a difficult attack trends. In this case, the F1-scores for the Linear and RBF did result in 96.65/99.24, respectively, indicating that SVM also proved to be an effective algorithm choice in this example. Finally, the study either propose new algorithms or improve existing algorithms for SVM and demonstrate their effectiveness in the area of network security. In conclusion, this study reinforces the importance of SVM as an effective instrument in the detection of attacks on network traffic and strengthening cyber defense capabilities.

关键词： Support vector machines Training Accuracy Scalability Intrusion detection Telecommunication traffic Network security Servers Kernel Testing

来源：评论

学校读者我要写书评

暂无评论

EyeUnderstand: Dashboard for Gaze and Deep-Learning Driven Comprehension Estimation in Online Lectures

引用

IEEE Access 2025年 13卷 102220-102233页

作者： Ko Watanabe Gitesh Gund Jayasankar Santhosh Haruka Sakagami Yuki Matsuda Andreas Dengel Shoya Ishimaru Smart Data and Knowledge Services German Research Center for Artificial Intelligence (DFKI) GmbH Kaiserslautern Germany Department of Computer Science RPTU Kaiserslautern-Landau Kaiserslautern Germany Graduate School of Science and Technology Nara Institute of Science and Technology (NAIST) Nara Japan Faculty of Environmental Life Natural Science and Technology Okayama University Okayama Japan Graduate School of Informatics Osaka Metropolitan University Sakai Japan

Online videos are a potent tool for educators to disseminate knowledge widely to diverse student audiences. However, collecting student feedback remains a significant challenge for lecturers, particularly in the absence of feedback. Understanding students’ subjective comprehension levels during online video lectures with sensor technology is yet to be thoroughly researched. This study uses eye-tracking technology to predict self-reported comprehension levels during video lectures. We recruited 20 participants from Germany and Japan who were invited to watch 50-minute lecture videos in three domains. The participants self-annotate the time segment in each lecture video where they dropout using open-source LabelStudio and answer the survey. We applied Long-Short-Term Memory (LSTM) to the preprocessed dataset and achieved an F1 Score of 0.886 for predicting binary self-annotated comprehension levels. We also introduce EyeUnderstand, the web-based application for visualizing the results of the comprehension estimation. We recruited 28 participants for the user study. As a result, 89.3% of the students and 92.9% of the lecturers confirmed that our application is practical.

关键词： Videos Estimation Gaze tracking Real-time systems Temperature measurement Education data visualization Vocabulary Surveys Pupils

来源：评论

学校读者我要写书评

暂无评论

Population-Based Incremental Learning for Effective IoT Service Composition with Replication

Population-Based Incremental Learning for Effective IoT Serv...

引用

Software Services Engineering (SSE) IEEE International Conference on

作者： Fengyang Sun Gang Chen Hui Ma Sven Hartmann Centre for Data Science and Artificial Intelligence & School of Engineering and Computer Science Victoria University of Wellington Wellington New Zealand Department of Informatics Clausthal University of Technology Clausthal-zellerfeld Germany

ISBN: (数字)9798350368512

ISBN: (纸本)9798350368529

Internet of Things Service Composition (SCIoT) aims to find the best composite IoT service to fulfil users' requirements. Given the NP-hard complexity of SCIoT, Evolutionary Computation methods, especially Estimation of Distribution Algorithms (EDAs), have received increasing attention to solve SCIoT problems. As one of the most popular EDA methods, Population-Based Incremental Learning (PBIL) has demonstrated its strong competency in optimising composed services in SCIoT. However, conventional PBIL does not explicitly utilize problem knowledge such as QoS and service replication, limiting its effectiveness for IoT service composition. In this paper, we propose a new PBIL based approach, named Population-Based Incremental Learning to Improve Service Composition (PBILISC), to solve the SCIoT problem. Different from traditional PBIL, PBILISC seamlessly integrates PBIL with QoS-aware local search (QLS) to effectively handle replicated services in the SCIoT problem. Specifically, PBILISC evolves a series of populations of solutions jointly through PBIL and QLS. PBIL leverages a probability distribution for population updates, while QLS focuses on improving the best evolved solution by searching promising neighboring solutions under the guidance of QoS. Experimental results show that PBILISC can outperform PBIL and several state-of-the-art methods on multiple benchmark SCIoT problems.

关键词： Incremental learning Limiting Software algorithms Quality of service Evolutionary computation Benchmark testing Search problems

来源：评论

学校读者我要写书评

暂无评论

DarkSAM: fooling segment anything model to segment nothing 24

DarkSAM: fooling segment anything model to segment nothing

引用

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Ziqi Zhou Yufei Song Minghui Li Shengshan Hu Xianlong Wang Leo Yu Zhang Dezhong Yao Hai Jin National Engineering Research Center for Big Data Technology and System and Services Computing Technology and System Lab and Cluster and Grid Computing Lab and School of Computer Science and Technology Huazhong University of Science and Technology School of Cyber Science and Engineering Huazhong University of Science and Technology School of Software Engineering Huazhong University of Science and Technology National Engineering Research Center for Big Data Technology and System and Services Computing Technology and System Lab and Hubei Engineering Research Center on Big Data Security and Hubei Key Laboratory of Distributed System Security and School of Cyber Science and Engineering Huazhong University of Science and Technology School of Information and Communication Technology Griffith University

ISBN: (纸本)9798331314385

Segment Anything Model (SAM) has recently gained much attention for its outstanding generalization to unseen data and tasks. Despite its promising prospect, the vulnerabilities of SAM, especially to universal adversarial perturbation (UAP) have not been thoroughly investigated yet. In this paper, we propose DarkSAM, the first prompt-free universal attack framework against SAM, including a semantic decoupling-based spatial attack and a texture distortion-based frequency attack. We first divide the output of SAM into foreground and background. Then, we design a shadow target strategy to obtain the semantic blueprint of the image as the attack target. DarkSAM is dedicated to fooling SAM by extracting and destroying crucial object features from images in both spatial and frequency domains. In the spatial domain, we disrupt the semantics of both the foreground and background in the image to confuse SAM. In the frequency domain, we further enhance the attack effectiveness by distorting the high-frequency components (i.e., texture information) of the image. Consequently, with a single UAP, DarkSAM renders SAM incapable of segmenting objects across diverse images with varying prompts. Experimental results on four datasets for SAM and its two variant models demonstrate the powerful attack capability and transferability of DarkSAM. Our codes are available at: https://***/CGCL-codes/DarkSAM.

关键词：

来源：评论

学校读者我要写书评

暂无评论

When Can Transformers Count to n?

arXiv

引用

arXiv 2024年

作者： Yehudai, Gilad Kaplan, Haim Ghandeharioun, Asma Geva, Mor Globerson, Amir Center for Data Science New York University United States Google Research United States Blavatnik School of Computer Science Tel Aviv University Google DeepMind United Kingdom

Large language models based on the transformer architecture can solve highly complex tasks. But are there simple tasks that such models cannot solve? Here we focus on very simple counting tasks, that involve counting how many times tokens in the vocabulary appeared in a string. We show that if the dimension of the transformer state is linear in the context length, this task can be solved. However, the solution we propose does not scale beyond this limit, and we provide theoretical arguments for why it is likely impossible for a size-limited transformer to implement this task. Our empirical results demonstrate the same phase-transition in performance, as anticipated by the theoretical argument. Our results demonstrate the importance of understanding how transformers can solve simple tasks. Copyright © 2024, The Authors. All rights reserved.

关键词： Distribution transformers

来源：评论

学校读者我要写书评

暂无评论

Generating Targeted Universal Adversarial Perturbation against Automatic Speech Recognition via Phoneme Tailoring

Generating Targeted Universal Adversarial Perturbation again...

引用

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

作者： Yujun Zhang Yanqu Chen Jiakai Wang Jin Hu Renshuai Tao Xianglong Liu State Key Laboratory of Complex & Critical Software Environment Beihang University China School of Computer Science and Engineering Beihang University China College of Computer Science Beijing University of Technology China Zhongguancun Laboratory China School of Computer and Information Technology Beijing Jiaotong University China Institute of Data Space Hefei Comprehensive National Science Center China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

There is a growing concern about adversarial attacks against automatic speech recognition (ASR) systems. Although research into targeted universal adversarial examples (AEs) has progressed, current methods are constrained by inefficient exploitation of audio features, demonstrating insufficient attack ability and robustness in the physical world. To solve this problem, we propose a phoneme-tailored attack (PTA) to improve the quality of the generated AEs. Specifically, to improve attack ability, we propose a Diverse Audio Composition Enrichment method, which enhances the utilization of audio features through phoneme-level slicing and recombination. To adapt AEs to complex environments, we propose a Natural Noise Pattern Guidance method to align AEs with natural noise patterns to improve their robustness. Experiments show that our method achieves an average accuracy of more than 72.34% and 98% with and without a norm constraint, and also demonstrates excellent performance in terms of generalization across datasets and resilience to MP3 compression.

关键词： Accuracy Target recognition Perturbation methods Noise Robustness Acoustics Speech processing Digital audio players Automatic speech recognition Resilience

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：