检索结果-内蒙古大学图书馆

14th ieee Pacific visualization symposium (ieee PacificVis)

作者： Zhou, Youjia Chalapathi, Nithin Rathore, Archit Zhao, Yaodong Wang, Bei Univ Utah Sci Comp & Imaging SCI Inst Salt Lake City UT 84112 USA

ISBN: (纸本)9781665439312

The mapper algorithm is a popular tool from topological data analysis for extracting topological summaries of high-dimensional datasets. In this paper, we present Mapper Interactive, a web-based framework for the interactive analysis and visualization of high-dimensional point cloud data. It implements the mapper algorithm in an interactive, scalable, and easily extendable way, thus supporting practical data analysis. In particular, its command-line API can compute mapper graphs for 1 million points of 256 dimensions in about 3 minutes (4 times faster than the vanilla implementation). Its visual interface allows on-the-fly computation and manipulation of the mapper graph based on user-specified parameters and supports the addition of new analysis modules with a few lines of code. Mapper Interactive makes the mapper algorithm accessible to nonspecialists and accelerates topological analytics workflows.

关键词： High Dimensional data Scalability Issues Topology based Techniques

来源：评论

学校读者我要写书评

暂无评论

Bandwidth Characterization of DeepSpeed on Distributed large Language Model Training

Bandwidth Characterization of DeepSpeed on Distributed Large...

引用

2024 ieee International symposium on Performance analysis of Systems and Software, ISPASS 2024

作者： Hanindhito, Bagus Patel, Bhavesh John, Lizy K. The University of Texas at Austin AustinTX United States Dell Technologies Round RockTX United States

ISBN: (纸本)9798350376388

The exponential growth of the training dataset and the size of the large language model (LLM) significantly outpaces the incremental memory capacity increase in the graphics pro-cessing units (GPUs). Thousands of GPUs are needed to handle state-of-the-art models, which require building an expensive AI GPU cluster that is out of reach for most researchers. This not only makes the cost to train the model more costly but also signifies the environmental impact. To improve the efficiency and scalability of existing infrastructure to handle increasingly demanding training tasks, Microsoft released DeepSpeed, an open-source optimization library for PyTorch that can easily be integrated into existing training flow with minimal code changes. This paper presents a comprehensive third-party evaluation of DeepSpeed for training GPT-2-like LLM on mainstream GPU clusters that are more accessible to everyone. The evaluation includes memory usage analysis and bandwidth characterization in addition to the achieved model size and the attained compute throughput to help compare horizontal and vertical scaling. First, we examine the DeepSpeed ZeRO in single- and dual-node training against the popular distributed training libraries: PyTorch Distributed data-Parallel (DDP) with data parallelism and Megatron-LM with data and model parallelism. While DDP achieves higher throughput due to less communication, the model size is limited to a single GPU memory capacity. In single-node training, Megatron-LM can fit a 4x larger model than the DDP, while ZeRO can handle a model with 0.8x-l.2x size of the Megatron-LM. Both Megatron-LM and ZeRO are reasonably competitive in terms of throughput. However, in dual-node training, Megatron- Lmsees a significant drop in throughput due to the excessive inter-node communication, achieving only 25 %-30 % of the throughput offered by ZeRO. Secondly, we evaluate ZeRO-Offload to consolidate multi-node training into single-node. With CPU offloading, ZeRO-Offloa

关键词： Bandwidth

来源：评论

学校读者我要写书评

暂无评论

Self-Prompt Guided Image Outpainting Model for Captions Absence in Social Scenes

引用

ieee TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS 2025年

作者： Zhang, Zongyan Chen, C. L. Philip Weng, Haohan Zhang, Tong South China Univ Technol Sch Comp Sci & Engn Guangdong Prov Key Lab AI Large Model & Intelligen Guangzhou 510006 Peoples R China Pazhou Lab Guangzhou 510335 Peoples R China Minist Educ Hlth Intelligent Percept & Paralleled Engn Res Ctr Guangzhou 510006 Peoples R China

The limitations of acquisition equipment often result in scene image data of limited size, posing a challenge for comprehensive analysis of social image datasets. Advances in generative models have introduced image outpainting techniques that expand the size of acquired social scene images, thereby enhancing the value of social image data. Stable diffusion (SD), which benefits from the guidance of caption prompts, shows excellent performance in image outpainting. However, its heavy reliance on manual prompts leads to a significant drawback: a decrease in the quality of generated images without prompts. To overcome this challenge, we propose a novel self-prompt diffusion model for image outpainting that extrapolates images based on the semantics of the source image, thereby removing the dependence on manual prompts. Specifically, we design a prompt autoencoder that uses an autoregressive transformer to map prompt embeddings into their semantic space, facilitating the construction of a semantic decoder. The semantic decoder and prompt embeddings are then cooptimized within the proposed prompt embedding network, allowing the mapping of image features to the stable diffusion prompt embeddings. Furthermore, by exploiting the inherent generative capabilities of diffusion models, we introduce a seam line regeneration mechanism to address the common problem of seam lines when splicing input and generated images. Comparative experiments on the Places2 and COCO datasets show that our method outperforms current state-of-the-art approaches on visual quality metrics and is adaptable to the stable diffusion model without additional fine-tuning.

关键词： Semantics Diffusion models visualization Image restoration Feature extraction Training Text to image Adaptation models Transformers Manuals Diffusion model generative model image process image outpainting

来源：评论

学校读者我要写书评

暂无评论

A Big data, Bigger Impact: A Comprehensive Review of Machine Learning Advancements 4

A Big Data, Bigger Impact: A Comprehensive Review of Machine...

引用

4th International Conference on Electrical Engineering and Computer Science, ICECOS 2024

作者： Dlshad Hamad Ameen, Didar Wahhab Kareem, Shahab Baez Hasan, Sarhad Soran University Faculty of Science Computer Science Department Soran Kurdistan Iraq Engineering Erbil Polytechnic University Erbil Department of Information System Kurdistan Iraq

ISBN: (纸本)9798350368253

The exponential growth of data driven by the internet has necessitated effective extraction of insights, with big data and machine learning standing as pivotal tools. This paper aims to provide insights into the evolving landscape of machine learning and big data analytics through a comprehensive literature review. It seeks to underscore the key advancements and intersections between these domains and their transformative applications across industries. Specifically, the study will delve into the foundational theories that underpin machine learning algorithms, data collection and preprocessing techniques involved in big data analysis, and the prominent challenges associated with extracting insights from large and complex datasets. It will further explore major studies that exemplify the integration of machine learning methodologies and big data technologies for addressing real-world problems. The study will also examine visualization methods that facilitate comprehending insights gleaned from big data. Finally, it will discuss future research directions and highlight best practices for leveraging machine learning and big data analytics in decision-making processes. © 2024 ieee.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

RESEARCH PROGRESS ON WIRELESS SENSOR NETWORKS FOR MECHANICAL VIBRATION MONITORING: A CITESPACE visualization analysis APPROACH 10

RESEARCH PROGRESS ON WIRELESS SENSOR NETWORKS FOR MECHANICAL...

引用

10th International symposium on Test Automation and Instrumentation, ISTAI 2024

作者： Zha, Zhendong Jiang, Zhanglei Li, Zihan Liu, Xiuli Wu, Guoxin The Ministry of Education Key Laboratory of Modern Measurement and Control Technology Beijing Information Science & Technology University Beijing100192 China

ISBN: (纸本)9781837242672

The application of wireless sensor networks (WSNs) in mechanical vibration monitoring is pivotal for achieving efficient monitoring within the realms of industrial automation and intelligent manufacturing, playing a significant role in the digital transformation and upgrading of China's manufacturing industry. This paper provides a comprehensive overview of the major technological advancements made by scholars and institutions globally over the past sixteen years in the domain of mechanical vibration monitoring using WSNs. Subsequently, employing CiteSpace for knowledge mapping and visualization, a multifaceted statistical analysis is conducted on over four hundred publications from both domestic and international sources. Through clustering analysis, keyword co-occurrence, and other analytical techniques, the paper delineates the multidimensional evolution of research hotspots and the dynamics of author collaboration networks over time. Building on this foundation, the paper delves into the cutting-edge trends of the field, emphasizing the critical importance of establishing efficient energy management systems and leveraging machine learning and artificial intelligence to process large-scale sensor data. These strategies are essential for optimizing system performance and propelling the development of WSNs towards miniaturization, integration, multifunctionality, systematization, and networking. The synergistic effects of these measures are anticipated to provide robust support for the intelligent upgrade of the manufacturing industry, fostering its progression towards greater efficiency and intelligence. © The Institution of Engineering & Technology 2024.

关键词： Information management

来源：评论

学校读者我要写书评

暂无评论

Free-form Conversation with Human and Symbolic Avatars in Mixed Reality 22

Free-form Conversation with Human and Symbolic Avatars in Mi...

引用

22nd ieee International symposium on Mixed and Augmented Reality (ISMAR)

作者： Zhu, Jiarui Kumaran, Radha Xu, Chengyuan Hollerer, Tobias Univ Calif Santa Barbara Santa Barbara CA 93106 USA

ISBN: (纸本)9798350328387

The integration of large language models and mixed reality technologies has enabled users to engage in free-form conversations with virtual agents across different "realities". However, if and how the agent's visual representation, especially when combined with mixed reality environments, will affect the conversation content or user experience is not yet fully understood. In this work, we design and conduct a user study involving two types of visual representations (a human avatar and a symbolic avatar) and two mixed reality environments (virtual reality and augmented reality), facilitating a free-form conversation experience with GPT-3 powered agents. We found evidence that the use of virtual or augmented realities can influence conversation content. Users chatting with avatars in virtual reality made significantly more references to the location or the space, suggesting they tended to perceive conversations as occurring in the agent's space, whereas the physical AR environment was perhaps more perceived as the user's space. Conversations with the human avatar improve user recall of the conversation, even though there is no evidence of increased information extracted during the conversation. These observations and our analysis of post-study questionnaires suggest that human avatars can positively impact user memory and experience. We hope our findings and the open-source implementation will help facilitate future research on free-form conversational agents in mixed reality.

关键词： Artificial intelligence (AI) large language models (LLMs) mixed reality Human-centered computing Empirical studies in HCI Computing methodologies Mixed / augmented reality Computing methodologies Virtual reality

来源：评论

学校读者我要写书评

暂无评论

RLBrowse: Generating Realistic Packet Traces with Reinforcement Learning

RLBrowse: Generating Realistic Packet Traces with Reinforcem...

引用

ieee/IFIP Network Operations and Management symposium

作者： Griessel, Alexander Stephan, Maximilian Mieth, Martin Kellerer, Wolfgang Kraemer, Patrick Tech Univ Munich Munich Germany Ipoque GmbH Leipzig Germany

ISBN: (纸本)9781665406017

Automated Web Browsing tools, such as Selenium and headless browsers, are used to collect traffic traces from networked applications, with which statistical models describing the traffic are obtained. However, we show that traces from Selenium and headless browsers have markedly different traffic characteristics than human generated traces, with potential impact on the quality of the obtained models. To overcome this limitation, we propose RLBRowsE, an automated web automation framework that imitates human browsing habits by separating web automation from the browser using reinforcement learning. By separating the browser and automation tool, RLBRowsE improves on 9 out of the 13 traffic trace features tested. The distribution of packet sizes in a trace improves the most, with a nearly 400 % improvement. We test RLBRowsE by collecting a corpus of network packet traces on a set of human-navigated website browsing sessions, and by RLBRowsE and Selenium. In the subsequent analysis, we identify key differences in the resulting packet traces.

关键词： visualization Automation Navigation Reinforcement learning data collection data models Browsers

来源：评论

学校读者我要写书评

暂无评论

Monte Carlo method based model for augmenting data towards lubricant oil state analysis in heavy machine industry

Monte Carlo method based model for augmenting data towards l...

引用

2023 ieee Applied Sensing Conference, APSCON 2023

作者： Chatterjee, Subhasri Gorey, Abhijeet Gain, Supriya Sinharay, Arijit Bhaumik, Chirabrata Chakravarty, Tapas Pal, Arpan Tcs Research India

ISBN: (纸本)9781665461634

Monitoring the health of lubricant oil used in a machine is one of the most effective ways to predict the machine condition. A non-invasive non-destructive monitoring approach may be used for a multiparametric causal analysis of the oil state. However, the analytical solution for the multiple contaminant determination in the oil is extremely challenging, which can be solved using a machine learning (ML) based classification approach. To train such an ML model needs a large number of datasets that is difficult to obtain experimentally. A computational model resembling the experimental system may be useful in this context. In this paper, a computational model replicating a non-invasive technique, namely, photoacoustic (PA) sensing to investigate lubricant oil is explored through a set of simulations followed by experimental validation and machine learning model-based data analysis. The effect of the two dominant causes of lubricant oil degradation, namely, soot particle and water has been investigated through simulation and experiment;and the ML analysis has been executed to classify the data into four classes depicting four states of the machine health. © 2023 ieee.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

visualization of Dynamic Network Evolution With Quantification of Node Attributes

引用

ieee TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING 2021年第3期8卷 2316-2325页

作者： Kim, Minseo Korea Univ Sejong Campus Cultural Heritage & Smart Technol Convergence Edu Sejong South Korea

The effective visual exploration of dynamic networks has been one of the toughest challenges and an unsolved problem;however, it is very important to understand network evolution. Although many developments have been achieved in modeling evolutionary networks, the closely related task of visualizing continues to remain a major concern. Therefore, in this study, quantitative analysis is used to assign node attributes in the network topology, and then, the evolutionary process of networks is analyzed. By fixing the position of nodes, the possibility of a stationary shape of the network is suggested, and a more intuitive and comprehensive explanation of the enumeration of the types for the evolutionary process is provided. Further, a large amount of information is presented in this study in an extremely economical and accessible way by incorporating a circular layout and evolution laws, which offers a new approach for the estimation and evaluation of network evolution. Finally, this three-pronged approach-network analysis, quantitative method, and topological modeling-is expected to provide a revelatory insight into the principle of network evolution.

关键词： data visualization data models Computational modeling visualization Network topology Analytical models Painting Art Communication Networking and Information Technology data and Knowledge visualization Evolution Image Processing and Computer Vision Simulation Modeling and visualization

来源：评论

学校读者我要写书评

暂无评论

Electrical Equipment Fault Diagnosis: A Technique Combining Fuzzy Logic and large Language Models

Electrical Equipment Fault Diagnosis: A Technique Combining ...

引用

2023 ieee International symposium on Product Compliance Engineering - Asia 2023, ISPCE-AS 2023

作者： Xu, Tao Tang, Xue-Song State Grid Shanghai Electric Power Company Ultra High Voltage Division Shanghai200063 China Donghua University Shanghai201620 China

ISBN: (纸本)9798350361209

Diagnosing faults in electrical equipment is a complex task that requires precise and adaptive decision-making. Traditional methodologies often fall short in handling the inherent uncertainties and vagueness associated with such diagnosis. In this paper, we introduce a novel technical approach that synergistically combines fuzzy logic with large language models, specifically ChatGLM2-6B. Fuzzy logic provides a structured framework to deal with imprecise information, while the language model offers a deep understanding of textual data related to faults. By taking local electrical text descriptions as input, we harness the power of fuzzy inference systems and the language model to generate comprehensive and accurate fault diagnosis results. This innovative fusion promises a more robust and adaptive fault diagnosis mechanism, ensuring safer and more efficient electrical infrastructure operations. © 2023 ieee.

关键词： Failure analysis

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：