检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Bi, Haoyang Ma, Haiping Huang, Zhenya Yin, Yu Liu, Qi Chen, Enhong Su, Yu Wang, Shijin Anhui Province Key Laboratory of Big Data Analysis and Application School of Computer Science and Technology University of Science and Technology of China China Anhui University China IFLYTEK Research

Computerized Adaptive Testing (CAT) is emerging as a promising testing application in many scenarios, such as education, game and recruitment, which targets at diagnosing the knowledge mastery levels of examinees on required concepts. It shows the advantage of tailoring a personalized testing procedure for each examinee, which selects questions step by step, depending on her performance. While there are many efforts on developing CAT systems, existing solutions generally follow an inflexible model-specific fashion. That is, they need to observe a specific cognitive model which can estimate examinee’s knowledge levels and design the selection strategy according to the model estimation. In this paper, we study a novel model-agnostic CAT problem, where we aim to propose a flexible framework that can adapt to different cognitive models. Meanwhile, this work also figures out CAT solution with addressing the problem of how to generate both high-quality and diverse questions simultaneously, which can give a comprehensive knowledge diagnosis for each examinee. Inspired by Active Learning, we propose a novel framework, namely Model-Agnostic Adaptive Testing (MAAT) for CAT solution, where we design three sophisticated modules including Quality Module, Diversity Module and Importance Module. Specifically, at one CAT selection step, Quality Module first quantifies the informativeness of questions and generates candidate subset with the highest quality. Then, Diversity Module selects one question at each step that maximizes the concept coverage. Additionally, we propose Importance Module to estimate the importance of concepts that optimizes the CAT selection. Under MAAT, we prove that the goal of maximizing both quality and diversity is NP-hard, but we provide efficient algorithms by exploiting the inherent submodular property. Extensive experimental results on two real-world datasets clearly demonstrate that our MAAT can support CAT with guaranteeing both quality and diversity

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

PseudoSC: A Binary Approximation to Stochastic Computing within Latent Operation-Space for Ultra-Lightweight on-Edge DNNs

PseudoSC: A Binary Approximation to Stochastic Computing wit...

引用

Design Automation Conference

作者： Jihe Wang Zhaoqing Wang Danghui Wang School of Computer Science Northwestern Polytechnical University Xi’an China Engineering Research Center of Embedded System Integration Ministry of Education Northwestern Polytechnical University Xi’an China National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Northwestern Polytechnical University Xi’an China

Recently, stochastic computing (SC) is increasingly popular in constructing MAC for on-edge DNNs benefiting from its outstanding energy-efficiency, including its adequate precision and gate-level operation. However, current SC-DNN systems always include a lot of costly SNGs/APCs for inevitably switches between binary and stochastic domains, which mortgages incongruous resources to pay the "bill" of the domain-switches and impedes highly-concurrent deployments. In this work, PseudoSC, a binary approximation to low-discrepancy SC, is proposed to totally remove the domain-switch for SNG/APC-free SC-DNNs. Its basic idea is to virtually re-arrange a couple of stochastic operands into a 2-D latent op-space, in which, original Monte Carlo sampling can be partitioned into three sub-ops, i.e., two fixed binary-ops and a fractal recursion. In theory, the recursion forms an isomorphic partition of the sampling repeated in smaller scales until the binary base-case achieved, as a result, a SC-op is well approximated only with binary-ops. Based on above theory, a multi-lane micro-architecture is designed to unroll the recursion within a few cycles and its advantages on hardware saving is verified under popular DNNs. The evaluation shows that the DNN-models with our schemes achieve 98.7% accuracy of the fixed-point implementations, which significantly outperform other SOTA methods. In addition, its reduced structure improves the power efficiency by 3.67 times on average.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Construction and Application of Knowledge Graph for Risk Identification Methods of Counterfeit Cigarettes and Wine

Construction and Application of Knowledge Graph for Risk Ide...

引用

Intelligent Communication and Networking (ICN), International Conference on

作者： Fulin Huang Jie Ning Lingjie Liu Ye Tian Jianghong Wu College of Computer Science and Technology Harbin Engineering University Harbin China Hangzhou Hikvision Digital Technology Co. Ltd Hangzhou China Environment and Food and Drug Crime Investigation Detachment Hangzhou Public Security Bureau Hangzhou China Hangzhou Institute of Technology Xidian University Hangzhou China Key Laboratory of Peace-Building Big Data of Zhejiang Province Hangzhou China

In response to the sharp increase in cases of counterfeit cigarettes and counterfeit alcohol in recent years, based on the application of knowledge graphs in social entities and social relationships, a knowledge graph for risk identification of counterfeit cigarettes and counterfeit alcohol was constructed and applied. Based on the historical cases of counterfeit cigarettes and counterfeit wines and the relevant groups, we process the data through ontology construction, knowledge extraction, and knowledge fusion, and build a knowledge map of relevant personnel for testing counterfeit wines. We use the TransH model to complete the constructed knowledge graph to obtain a more complete knowledge graph for risk identification of counterfeit cigarettes and counterfeit alcohol, and then use PageRank and Louvain community discovery algorithms to screen out communities where people with major suspicions gather, and control the system. Gangs selling counterfeit cigarettes and alcohol provide risk warning.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Ascl: Accelerating Semi-Supervised Learning Via Contrastive Learning

SSRN

引用

SSRN 2024年

作者： Liu, Haixiong Li, Zuoyong Wu, Jiawei Zeng, Kun Hu, Rong Zeng, Wei Fujian Provincial Key Laboratory of Big Data Mining and Applications School of Computer Science and Mathematics Fujian University of Technology Fuzhou350118 China Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Data Science Minjiang University Fuzhou350121 China School of Intelligent Systems Engineering Shenzhen Campus of Sun Yat-sen University Guangdong Shenzhen518107 China The Key Laboratory of Cognitive Computing and Intelligent Information Processing of Fujian Education Institutions Wuyi University Wuyishan354300 China School of Physics and Mechanical and Electrical Engineering Longyan University Longyan364012 China

SSL(Semi-supervised learning) is widely used in machine learning, which leverages labeled and unlabeled data to improve model performance. SSL aims to optimize class mutual information, but noisy pseudo-labels introduce false class information due to the scarcity of labels. Therefore, these algorithms often need significant training time to refine pseudo-labels for performance improvement iteratively. To tackle this challenge, we propose a novel plug-and-play method named ASCL(Accelerating Semi-Supervised Learning via Contrastive Learning). This method combines contrastive learning with uncertainty-based selection for performance improvement and accelerates the convergence of SSL algorithms. Contrastive learning initially emphasizes the mutual information between samples as a means to decrease dependence on pseudo-labels. Subsequently, it gradually turns to maximizing the mutual information between classes, aligning with the objective of semi-supervised learning. Uncertainty-based selection provides a robust mechanism for acquiring pseudo-labels. The combination of the contrastive learning module and the uncertainty-based selection module forms a virtuous cycle to improve the performance of the proposed model. Extensive experiments demonstrate that ASCL outperforms state-of-the-art methods in terms of both convergence efficiency and performance. In the experimental scenario where only one label is assigned per class in the CIFAR-10 dataset, the application of ASCL to Pseudo-label, UDA (Unsupervised data Augmentation for Consistency Training), and Fixmatch benefits substantial improvements in classification accuracy. Specifically, the results demonstrate notable improvements in respect of 16.32%, 6.9%, and 24.43% when compared to the original outcomes. Moreover, the required training time is reduced by almost 50%. © 2024, The Authors. All rights reserved.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Failure-Tolerant Task Offloading for Vehicular Fog Computing

Failure-Tolerant Task Offloading for Vehicular Fog Computing

引用

Symposia and Workshops on Ubiquitous, Autonomic and Trusted Computing, UIC-ATC

作者： Chaogang Tang Huaming Wu Chunsheng Zhu Shuo Xiao Haifeng Jiang Dong Zeng School of Computer Science and Technology China University of Mining and Technology Xuzhou China The Center for Applied Mathematics Tianjin University Tianjin China College of Big Data and Internet Shenzhen Technology University Shenzhen Guangdong China School of Architecture and Design China University of Mining and Technology Xuzhou China

Vehicular fog computing (VFC) as one of the newly emergent computing paradigms has attracted widespread attention recently, which brings the computing and storage resources of cloud computing to the mobile terminal devices – smart vehicles. Despite the advantage of VFC, tasks are more susceptible to failures, owing to the limited computing capabilities of vehicles as well as the unstable wireless communication links during service provisioning. However, current works seldom pay attention to it. In this paper we strive to design failure-tolerant task offloading strategy in VFC, by considering different kinds of failures during task offloading. We also pose strict latency requirement on the response delay for each task in the optimization period. The goal of this paper is to minimize the average response latency for all the offloaded tasks in the optimization period by allocating the offloading requests to the fog vehicles. The optimization problem is formulated as one variant of multiple knapsack problem that is NP-hard. To solve this problem in polynomial time, we put forward a heuristic greedy algorithm to solve it. The extensive simulation is conducted to validate its efficiency and effectiveness and the results show its advantages compared to the baseline approaches.

关键词：

来源：评论

学校读者我要写书评

暂无评论

M3CS: Multi-Target Masked Point Modeling with Learnable Codebook and Siamese Decoders

引用

IEEE Transactions on Circuits and Systems for Video technology 2025年

作者： Qiu, Qibo Yang, Honghui Jiang, Jian Zhang, Shun Ying, Haochao Gao, Haiming Wang, Wenxiao He, Xiaofei Zhejiang University State Key Lab of CAD&CG China Research & Innovation Institute China Zhejiang Lab China School of Public Health Zhejiang University Department of Big Data in Health Science China School of Medicine Zhejiang University Department of Breast Surgery and Oncology Second Affiliated Hospital China Zhejiang University ZJU-Hangzhou Global Scientific and Technological Innovation Center China Zhejiang University School of Software Technology China

Masked point modeling has become a promising scheme of self-supervised pre-training for point clouds. Existing methods reconstruct either the masked points or related features as the objective of pre-training. However, considering the diversity of downstream tasks, it is necessary for the model to have both low- and high-level representation modeling capabilities during pre-training. It enables the capture of both geometric details and semantic contexts. To this end, M3CS is proposed to endow the model with the above abilities. Specifically, with the masked point cloud as input, M3CS introduces two decoders to reconstruct masked representations and the masked points simultaneously. While an extra decoder doubles parameters for the decoding process and may lead to overfitting, we propose siamese decoders to keep the number of learnable parameters unchanged. Further, we propose an online codebook projecting continuous tokens into discrete ones before reconstructing masked points. In such a way, we can compel the decoder to take effect through the combinations of tokens rather than remembering each token. Comprehensive experiments show that M3CS achieves superior performance across both classification and segmentation tasks, outperforming existing methods that are also single-modality and single-scale. © 2025 IEEE. All rights reserved.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

data Compression Algorithms for Bridge Health Monitoring

引用

science technology and Engineering 2025年第8期25卷 3304-3315页

作者： Chen, Ken Zhong, Ai-Ping Wen, Yu-Xuan Yang, Yang Li, Wei Zeng, Shan Wang, Jun Tan, Qu-Shan Yang, Liu Sichuan Digital Transpotation Tech Co. Ltd. Chengdu610095 China National Engineering Laboratory of Integrated Transportation Big Data Application Technology Southwest Jiaotong University Chengdu611756 China School of Information Science and Technology Southwest Jiaotong University Chengdu611756 China

The bridge health monitoring system based on sensor data acquisition has become standard for new bridge construction. However, this scenario presents challenges due to the massive volume of monitoring data that is difficult to store. Therefore, focusing on the time-series characteristics of bridge monitoring data, compression schemes were explored for bridge monitoring data. Differential compression was investigated based on the arithmetic progression properties of bridge monitoring timestamps and floating-point exclusive OR(XOR) compression based on the low frequency of changes in monitoring value data. Compared to the Gorilla time series database algorithm, the XOR compression method added control bits to avoid degradation of compression results. Experimental analysis reveals that both algorithms exhibit varying degrees of superiority over common compressors. The differential compression of timestamp sequences demonstrates superior compression rates compared to common compressors, achieving a compression rate of 0. 015 6 for timestamp sequences that conform to arithmetic progression characteristics, approaching the compression limit value. Compression and decompression speeds are above average, and the method is insensitive to monitoring type. On the other hand, the XOR compression method performs well on datasets with low frequency of change, achieving compression rates of 0. 302 8 for bridge data and 0. 662 8 for non-bridge data, indicating sensitivity of the XOR compression method to monitoring type. In practical applications of bridge monitoring, suitable compression storage schemes can be selected based on the characteristics of the bridge monitoring dataset. © 2025 Beijing Kexue Jishu yu Gongcheng Zazhishe. All rights reserved.

关键词： Digital arithmetic

来源：评论

学校读者我要写书评

暂无评论

Migration Privacy Protection Based on Scheduling Algorithm for Online Car-Hailing 7th

Migration Privacy Protection Based on Scheduling Algorithm f...

引用

7th International Symposium on Security and Privacy in Social Networks and big data, SocialSec 2021

作者： Ding, Qian Zhang, Jing Lin, Liwei Xu, Zhiping Wang, Yichun School of Computer Science and Mathematics Fujian University of Technology Fujian Fuzhou350118 China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fujian Fuzhou350118 China Shanghai Oriental Pearl Digital TV Company Limited 1 Century Avenue Pudong District Shanghai200120 China

ISBN: (纸本)9789811679124

In the era of rapid development of mobile Internet technology and big data technology, people enjoy more convenient travel because of Online Car-hailing. At the same time, the location privacy of drivers and passengers has also been leaked due to the use of Online Car-hailing. The traditional privacy preserving encryption technology for Online Car-hailing has a large amount of computation and low efficiency. In order to solve these problems, the Migration Privacy Protection based on Scheduling Algorithm (MPSA) for Online car-hailing is proposed in this paper. Firstly, the road network model is established through the road network matching technology. Secondly, the shortest path between the passenger and the car is calculated through A* algorithm and Dijkstra algorithm. Finally, the location privacy protection between the driver and the passenger is realized through the offset technology. In this paper, a scientific experiment is carried out on real data to verify the superiority of MPSA in privacy protection effect and operation efficiency. © 2021, Springer Nature Singapore Pte Ltd.

关键词： Location privacy protection Migration technology Online Car-hailing Road network matching

来源：评论

学校读者我要写书评

暂无评论

Multi-level Relation Learning for Cross-domain Few-shot Hyperspectral Image Classification

arXiv

引用

arXiv 2023年

作者： Liu, Chun Yang, Longwei Li, Zheng Yang, Wei Han, Zhigang Guo, Jianzhong Yu, Junyong School of Computer and Information Engineering Henan University Henan Zhengzhou450046 China Henan Key Laboratory of Big Data Analysis and Processing Henan University Henan Zhengzhou450046 China Henan Engineering Laboratory of Spatial Information Processing Henan University Henan Zhengzhou450046 China Henan Industrial Technology Academy of Spatio-Temporal Big Data Henan University Henan Zhengzhou450046 China College of Geography and Environmental Science Henan University Henan Zhengzhou450046 China Henan DeFan High-tech Software Co. Ltd Henan Zhengzhou450046 China

Cross-domain few-shot hyperspectral image classification focuses on learning prior knowledge from a large number of labeled samples from source domains and then transferring the knowledge to the tasks which contain few labeled samples in target domains. Following the metric-based manner, many current methods first extract the features of the query and support samples, and then directly predict the classes of query samples according to their distance to the support samples or prototypes. The relations between samples have not been fully explored and utilized. Different from current works, this paper proposes to learn sample relations on different levels and take them into the model learning process, to improve the cross-domain few-shot hyperspectral image classification. Building on current method of"Deep Cross-Domain Few-Shot Learning for Hyperspectral Image Classification" which adopts a domain discriminator to deal with domain-level distribution difference, the proposed method applies contrastive learning to learn the class-level sample relations to obtain more discriminable sample features. In addition, it adopts a transformer based cross-attention learning module to learn the set-level sample relations and acquire the attention from query samples to support samples. Our experimental results have demonstrated the contribution of the multi-level relation learning mechanism for few-shot hyperspectral image classification when compared with the state of the art methods. All the codes are available at github https://***/HENULWY/STBDIP. © 2023, CC BY.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Research on Prediction of Medical Training Effect Based on PSO-BP Neural Network

Research on Prediction of Medical Training Effect Based on P...

引用

IEEE International Conference on Orange Technologies (ICOT)

作者： Zilong Yin Xin Guo Shih-Pang Tseng Lujia Tang Ying Chen School of Information Science and Technology Sanda University Shanghai China School of Software and Big Data Changzhou College of Information Technology Changzhou China Emergency Department Xin Hua Hospital affiliated to Shanghai Jiao Tong University School of Medicine Shanghai China

ISBN: (纸本)9781665454209

Today, the world is still suffering from Coronavirus disease 2019(COVID-19) and other disasters. Therefore, it is critical to improve medical emergency professional training, and ensuring the training effect has become the top priority. As a result, this paper builds a Particle Swarm Optimization Back Propagation(PSO-BP) neural network model using training data from the National Disaster Life Support(NDLS) course to predict NDLS training outcomes. The PSO algorithm is used to calculate the initial weights of the BP network, and the model is then trained using error back propagation to obtain the predicted value of the training effect. When compared to the standard BP neural network prediction results, experimental analysis shows that the prediction model's accuracy reaches 93.24 percentage, and the prediction accuracy is improved by 11.71 percentage. It is also better in terms of convergence speed, minimum error, global search ability, and learning smoothness. This approach is suitable for medical training effect prediction and additionally to assist the training providers in grasping trainees' learning effects in advance to improve training quality.

关键词： Training COVID-19 Training data Predictive models Prediction algorithms Social implications of technology Biological neural networks

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：