检索结果-内蒙古大学图书馆

Visual-semantic network: a visual and semantic enhanced model for gesture recognition

Visual Intelligence 2023年第1期1卷 71-83页

作者： Yizhe Wang Congqi Cao Yanning Zhang National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer ScienceNorthwestern Polytechnical UniversityXi’an710129China

Gesture recognition has attracted considerable attention and made encouraging progress in recent years due to its great potential in ***,the spatial and temporal modeling in gesture recognition is still a problem to be ***,existing works lack efficient temporal modeling and effective spatial attention *** efficiently model temporal information,wefirst propose a long-and short-term temporal shift module(LS-TSM)that models the long-term and short-term temporal information ***,we propose a spatial attention module(SAM)that focuses on where the change primarily occurs to obtain effective spatial attention *** addition,the semantic relationship among gestures is helpful in gesture ***,this is usually neglected by previous ***,we propose a label relation module(LRM)that takes full advantage of the relationship among classes based on their labels’semantic *** explore the best form of LRM,we design four different semantic reconstruction methods to incorporate the semantic relationship information into the class label’s semantic *** perform extensive ablation studies to analyze the best settings of each *** best form of LRM is utilized to build our visual-semantic network(VS Network),which achieves the state-of-the-art performance on two gesture datasets,i.e.,EgoGesture and NVGesture.

关键词： Gesture recognition Temporal modeling Spatial attention Semantic relationship

来源：评论

学校读者我要写书评

暂无评论

EvoChart: A Benchmark and a Self-Training Approach Towards Real-World Chart Understanding 39

EvoChart: A Benchmark and a Self-Training Approach Towards R...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Huang, Muye Lai, Han Zhang, Xinyu Wu, Wenjun Ma, Jie Zhang, Lingling Liu, Jun School of Computer Science and Technology Xi’an Jiaotong University China MOE KLINNS Lab Xi’an Jiaotong University China Shaanxi Province Key Laboratory of Big Data Knowledge Engineering China

ISBN: (纸本)157735897X

Chart understanding enables automated data analysis for humans, which requires models to achieve highly accurate visual comprehension. While existing Visual Language Models (VLMs) have shown progress in chart understanding, the lack of high-quality training data and comprehensive evaluation benchmarks hinders VLM chart comprehension. In this paper, we introduce EvoChart, a novel self-training method for generating synthetic chart data to enhance VLMs’ capabilities in real-world chart comprehension. We also propose EvoChart-QA, a noval benchmark for measuring models’ chart comprehension abilities in real-world scenarios. Specifically, EvoChart is a unique self-training data synthesis approach that simultaneously produces high-quality training corpus and a high-performance chart understanding model. EvoChart-QA consists of 650 distinct real-world charts collected from 140 different websites and 1,250 expert-curated questions that focus on chart understanding. Experimental results on various open-source and proprietary VLMs tested on EvoChart-QA demonstrate that even the best proprietary model, GPT-4o, achieves only 49.8% accuracy. Moreover, the EvoChart method significantly boosts the performance of open-source VLMs on real-world chart understanding tasks, achieving 54.2% accuracy on EvoChart-QA. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ContTune: Continuous Tuning by Conservative Bayesian Optimization for Distributed Stream data Processing Systems 49th

ContTune: Continuous Tuning by Conservative Bayesian Optimiz...

引用

49th International Conference on Very Large data Bases, VLDB 2023

作者： Lian, Jinqing Zhang, Xinyi Shao, Yingxia Pu, Zenglin Xiang, Qingfeng Li, Yawen Cui, Bin Beijing University of Posts and Telecommunications China School of Computer Science and National Engineering Laboratory for Big Data Analysis and Applications Peking University China

The past decade has seen rapid growth of distributed stream data processing systems. Under these systems, a stream application is realized as a Directed Acyclic Graph (DAG) of operators, where the level of parallelism of each operator has a substantial impact on its overall performance. However, finding optimal levels of parallelism remains challenging. Most existing methods are heavily coupled with the topological graph of operators, unable to efficiently tune under-provisioned jobs. They either insufficiently use previous tuning experience by treating successively tuning independently, or explore the configuration space aggressively, violating the Service Level Agreements (SLA). To address the above problems, we propose ContTune, a continuous tuning system for stream applications. It is equipped with a novel Big-small algorithm, in which the Big phase decouples the tuning from the topological graph by decomposing the job tuning problem into sub-problems that can be solved concurrently. We propose a conservative Bayesian Optimization (CBO) technique in the Small phase to speed up the tuning process by utilizing the previous observations. It leverages the state-of-the-art (SOTA) tuning method as conservative exploration to avoid SLA violations. Experimental results show that ContTune reduces up to 60.75% number of reconfigurations under synthetic workloads and up to 57.5% number of reconfigurations under real workloads, compared to the SOTA method DS2. © 2023, VLDB Endowment. All rights reserved.

关键词： Directed graphs

来源：评论

学校读者我要写书评

暂无评论

NLP-Based Analysis of Annual Reports: Asset Volatility Prediction and Portfolio Strategy Application 32

NLP-Based Analysis of Annual Reports: Asset Volatility Predi...

引用

32nd Irish Conference on Artificial Intelligence and Cognitive science, AICS 2024

作者： Li, Xiao Xu, Yang Yang, Linyi Zhang, Yue Dong, Ruihai Insight Centre for Data Analytics School of Computer Science University College Dublin Ireland School of Economics and Management Beihang University Beijing China Zhejiang University Zhejiang China School of Engineering Westlake University Zhejiang China

Leveraging recent developments in natural language processing (NLP), we constructed a prediction model using corporate financial annual reports to forecast the stock volatility indicator Beta (β), by analyzing risk discussions. The predicted Beta values were used to construct investment portfolios, whose market performance was then evaluated. Our research demonstrates that the Hierarchical Transformer-based model effectively captures complex risk information from annual reports, leading to improved returns in portfolio simulations. Our motivation arises from the need to better understand and process long, unstructured financial texts like annual reports, which contain crucial yet nuanced risk factors. By utilizing the hierarchical model, we aim to overcome traditional models’ limitations in handling such long documents, thereby improving the model’s understanding of both sentence-level and document-level contexts. The results highlight the potential of deep learning, particularly hierarchical models, in financial text prediction, and provide a novel perspective on asset management strategies. Compared to the S&P 500 benchmark, portfolios constructed using the predicted Beta values from our model achieved an average return increase of 21% over the same period. © 2024 Copyright for this paper by its authors.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Metamorphic Testing for Traffic Sign Detection and Recognition 23

Metamorphic Testing for Traffic Sign Detection and Recogniti...

引用

23rd IEEE International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023

作者： Long, Yuanyuan Fan, Yong Pan, Ya School of Computer Science and Technology of Southwest University of Science and Technology Mianyang China Sichuan Big Data and Intelligent System Engineering Technology Research Center Mianyang China

ISBN: (纸本)9798350359398

Traffic sign detection and recognition (TSDR) is a crucial technology to realize vehicle autonomous driving and maintain road safety. Misdetection or misclassification of traffic signs can lead to serious traffic accidents. Therefore, it is essential to ensure the correctness of traffic sign detection and recognition systems. However, most existing methods use labeled data, but it isn't easy to obtain labeled data. The metamorphic testing can reduce the dependence on the labeled data, but the current metamorphic relations do not consider the domain characteristics of traffic signs, so it is difficult to effectively detect traffic sign detection and recognition systems faults. To solve the above problems, this paper introduces metamorphic testing to verify traffic sign detection and recognition systems and designs and implements a tool, MTSDR, which provides a series of metamorphic relations in the traffic domain. It can generate many follow-up test images based on these metamorphic relations and automatically performs tests for evaluation, focusing on practical failure situations. The experimental evaluation was carried out on four advanced traffic sign detection and recognition models on open datasets. The results show that MTSDR can generate natural images and find thousands of traffic sign detection and recognition faults. Compared with the previous methods, our method has a higher fault detection capability, and the highest fault detection rate reaches more than 40%. In addition, it finds that retraining the traffic sign detection and recognition models using the images that found the fault can improve the model's performance, reduce its fault, and reduce the fault rate by 11.77% on average. © 2023 IEEE.

关键词： Traffic signs

来源：评论

学校读者我要写书评

暂无评论

Subgroup-Adaptive Network for Robust Virtual Patient-Specific Quality Assurance 20

Subgroup-Adaptive Network for Robust Virtual Patient-Specifi...

引用

20th International computer Conference on Wavelet Active Media Technology and Information Processing, ICCWAMTIP 2023

作者： Zeng, Xiaoyang Ahmed, Awais Tunio, Muhammad Hanif Chengxiao, Yan Hou, Mengshu Xi, Rui School of Computer Science and Engineering University of Electronic Science and Technology of China - UESTC Sichuan611731 China School of Big Data and Artificial Intelligence Chengdu Technological University Sichuan611730 China

ISBN: (纸本)9798350318982

Deep learning-based automatic patient-specific quality assurance (PSQA) alleviates medical resource pressure and ensures the safety of patient treatment plans by predicting the actual dosage difference distribution or metrics. However, deep learning is often affected by extremely unbalanced data distributions, and in clinical practice, the number of normal plans is always much larger than that of abnormal plans. On the other hand, the distribution of patients' lesion sites could be more balanced. The performance of models is often seriously biased in these subgroups, posing a challenge to the reliability of deep learning in the clinical application of PSQA. In this work, we propose a subgroup-adaptive network to address the problem of subgroup drift in PSQA. Our experiments achieved a state-of-the-art (SOTA) MAE of 0.69. Our proposed work can significantly improve the reliability of deep learning in the practical clinical application of PSQA. © 2023 IEEE.

关键词： Quality assurance

来源：评论

学校读者我要写书评

暂无评论

Multi-level Learning for Semi-supervised Multi-view Classification 11

Multi-level Learning for Semi-supervised Multi-view Classifi...

引用

11th International Conference on Advanced Cloud and Big data, CBD 2023

作者： Song, Na Yang, Shundong Cai, Zhiling Lin, Jian Huang, Sujia Putian University School of Mechanical Electrical and Information Engineering Putian China Hainan University School of Computer Science and Technology Haikou China Fujian Agriculture and Forestry University College of Computer and Information Science Fuzhou China Fuzhou University College of Computer and Data Science Fuzhou China

ISBN: (纸本)9798350345346

Multi-view semi-supervised classification primarily aims to enhance classification accuracy when dealing with limited labeled samples. Although existing methods have shown impressive performance, significant challenges still persist in efficiently propagating label information and capturing global relationship information within multi-view data. To address the aforementioned challenges, a Graph Convolutional Network (GCN)-based framework is presented, which explores the acquisition of multilevel representations, encompassing feature-level representations and the fusion of global structural associations. The state-of-the-art performance of the proposed model is unequivocally supported by a comprehensive set of empirical results. © 2023 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

A Framework for Multiclass Contour Visualization

引用

IEEE Transactions on Visualization and computer Graphics 2023年第1期29卷 353-362页

作者： Li, Sihang Yu, Jiacheng Li, Mingxuan Liu, Le Zhang, Xiaolong Luke Yuan, Xiaoru Laboratory of Machine Perception Ministry of Education School of AI Peking University China Center for Data Science Academy for Advanced Interdisciplinary Studies Peking University China National Engineering Laboratory for Big Data Analysis and Application Peking University China School of Computer Science Northwestern Polytechnical University China College of Information Sciences and Technology Pennsylvania State University United States

Multiclass contour visualization is often used to interpret complex data attributes in such fields as weather forecasting, computational fluid dynamics, and artificial intelligence. However, effective and accurate representations of underlying data patterns and correlations can be challenging in multiclass contour visualization, primarily due to the inevitable visual cluttering and occlusions when the number of classes is significant. To address this issue, visualization design must carefully choose design parameters to make visualization more comprehensible. With this goal in mind, we proposed a framework for multiclass contour visualization. The framework has two components: a set of four visualization design parameters, which are developed based on an extensive review of literature on contour visualization, and a declarative domain-specific language (DSL) for creating multiclass contour rendering, which enables a fast exploration of those design parameters. A task-oriented user study was conducted to assess how those design parameters affect users' interpretations of real-world data. The study results offered some suggestions on the value choices of design parameters in multiclass contour visualization. © 2022 IEEE.

关键词： data visualization

来源：评论

学校读者我要写书评

暂无评论

Research on the algorithm of searching building contour line by line segment based on Revit

Research on the algorithm of searching building contour line...

引用

2023 China Automation Congress, CAC 2023

作者： Zhong, Hui Wang, Wen School of Computer Science and Engineering Shenyang Jianzhu University Liaoning Province Big Data Management and Analysisline Laboratory of Urban Construction Shenyang China

ISBN: (纸本)9798350303759

The extraction of building contour line is an important computer-aided design work for the layout of building climbing frame walkway plates. So a study is conducted on the identification of building components and their characteristics for planar CAD drawings, and an algorithm for searching building contour line by line segment is proposed. Import CAD building plan data in Revit, read CAD drawing geometric information through Teigha class library, identify the walls forming the building frame, analyze the geometric features of the walls forming the building contour, the algorithm of searching the building contour line is proposed by line segment, which can effectively solve the problem of identifying the corner contour of wall break and inward depression, and can automatically extract the whole building contour from the plan CAD drawing data, laying the foundation for the optimized layout of the auxiliary climbing frame walkway plates. The experimental results show that the line segment search algorithm can effectively extract the contour of buildings from existing CAD drawing data, improving the efficiency of building climbing frame BIM modeling. © 2023 IEEE.

关键词： data mining

来源：评论

学校读者我要写书评

暂无评论

A Few-shot Classification Algorithm Based on Two-stream Frequency-Domain Information Networks 9

A Few-shot Classification Algorithm Based on Two-stream Freq...

引用

9th International Conference on Systems and Informatics, ICSAI 2023

作者： Hu, Siquan Jia, Jun Shi, Zhiguo Zhao, Jun University of Science and Technology Beijing School of Computer and Communication Engineering Beijing100083 China Beijing Big Data Center Beijing100101 China

ISBN: (纸本)9798350383706

In the domain of few-shot learning, where the scarcity of training data poses a significant challenge, this paper introduces an innovative approach. We present a few-shot classification algorithm that utilizes the Two-Stream Frequency Domain Information Network. This algorithm delves deep into the efficacy of training different frequency components of images in the context of few-shot classification. It integrates filtered and enhanced image frequency domain information as supplementary data to be adaptively fused with the original image data, enhancing and augmenting the available information. Experimental results establish that the few-shot classification model, employing the Two-Stream Frequency Domain Information Network, outperforms leading algorithms in the field of few-shot learning. This enhancement is evidenced by an accuracy improvement of 1.06% and 0.93% on the miniImageNet and tiredImageNet datasets, respectively, when compared to the Meta-Baseline, illustrating the efficacy of this specialized approach. © 2023 IEEE.

关键词： Frequency domain 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：