检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Zhan, Jingtao Xie, Xiaohui Mao, Jiaxin Liu, Yiqun Guo, Jiafeng Zhang, Min Ma, Shaoping Department of Computer Science and Technology Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China Beijing Key Laboratory of Big Data Management and Analysis Methods Gaoling School of Artificial Intelligence Renmin University of China Beijing100872 China CAS Key Lab of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China

A retrieval model should not only interpolate the training data but also extrapolate well to the queries that are different from the training data. While neural retrieval models have demonstrated impressive performance on ad-hoc search benchmarks, we still know little about how they perform in terms of interpolation and extrapolation. In this paper, we demonstrate the importance of separately evaluating the two capabilities of neural retrieval models. Firstly, we examine existing ad-hoc search benchmarks from the two perspectives. We investigate the distribution of training and test data and find a considerable overlap in query entities, query intent, and relevance labels. This finding implies that the evaluation on these test sets is biased toward interpolation and cannot accurately reflect the extrapolation capacity. Secondly, we propose a novel evaluation protocol to separately evaluate the interpolation and extrapolation performance on existing benchmark datasets. It resamples the training and test data based on query similarity and utilizes the resampled dataset for training and evaluation. Finally, we leverage the proposed evaluation protocol to comprehensively revisit a number of widely-adopted neural retrieval models. Results show models perform differently when moving from interpolation to extrapolation. For example, representation-based retrieval models perform almost as well as interaction-based retrieval models in terms of interpolation but not extrapolation. Therefore, it is necessary to separately evaluate both interpolation and extrapolation performance and the proposed resampling method serves as a simple yet effective evaluation tool for future IR studies. Copyright © 2022, The Authors. All rights reserved.

关键词： Extrapolation

来源：评论

学校读者我要写书评

暂无评论

Feature Aggregation network for Memory-Based Industrial Anomaly Detection

Feature Aggregation Network for Memory-Based Industrial Anom...

引用

IEEE International Conference on data science in Cyberspace (DSC)

作者： Min Li Delong Han Rui Shao Jinghui He Faculty of Data Science City University of Macau China Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350391367

ISBN: (纸本)9798350391374

In the process of large-scale industrial production, complex defects of different shapes and sizes can occur in industrial products and their components, and these defects seriously affect the performance of industrial products and their components, and being able to quickly and accurately identify defective industrial products and their components is an important task. Memory-based methods are usually used for anomaly detection, and these methods often rely on pre-trained models for feature extraction, but have high requirements on feature quality. In addition, patch-based methods mainly focus on local differences in feature embedding when detecting anomalies without explicitly exploiting broader contextual information, thus greatly affecting the robustness and accuracy of anomaly detection. Aiming at the above problems and requirements, this paper proposes a Feature Aggregation network for Memory-Based Industrial Anomaly Detection (FANet), which is based on the pre-extraction of anomalous features, attentional attention to key features, information aggregation of multi-level and multi-scale features, and the FANet provides advanced performance in industrial image anomaly detection and localisation by pre-extraction of anomalous features, attentional attention to key features, information aggregation of multi-layered and multi-scale features, and introduction of environmental information in the anomaly score computation to perform industrial anomaly detection tasks, and using core set subsampling to reduce storage requirements. On the MVTec AD, FANet has an image-level AUROC score of 99.1% and a pixel-level AUROC score of 98.2%. Similarly high detection results were achieved on the ViSA dataset, further demonstrating the robustness of FANet.

关键词： Training Runtime Shape Memory management Training data Production Feature extraction Robustness Anomaly detection Testing

来源：评论

学校读者我要写书评

暂无评论

Selection for Cheaper Amino Acids Drives Nucleotide Usage at the Start of Translation in Eukaryotic Genes

引用

Genomics, Proteomics & Bioinformatics 2021年第6期19卷 949-957页

作者： Na L.Gao Zilong He Qianhui Zhu Puzi Jiang Songnian Hu Wei-Hua Chen Key Laboratory of Molecular Biophysics of the Ministry of Education Hubei Key Laboratory of Bioinformatics and Molecular-imagingDepartment of Bioinformatics and Systems BiologyCollege of Life Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China Institute for Computer Science and Cluster of Excellence on Plant Sciences Heinrich Heine UniversityD-40225 DuesseldorfGermany CAS Key Laboratory of Genome Sciences and Information Beijing Institute of GenomicsChinese Academy of SciencesBeijing 100101China State Key Laboratory of Microbial Resources Institute of MicrobiologyChinese Academy of SciencesBeijing 100101China Beijing Advanced Innovation Center for Big Data-Based Precision Medicine Interdisciplinary Innovation Institute of Medicine and EngineeringBeihang UniversityBeijing 100191China University of Chinese Academy of Sciences Beijing 100049China

Coding regions have complex interactions among multiple selective forces,which are manifested as biases in nucleotide *** studies have revealed a decreasing GC gradient from the 5′-end to 3′-end of coding regions in various *** confirmed that this gradient is universal in eukaryotic genes,but the decrease only starts from the~25th *** trend is mostly found in nonsynonymous(ns)sites at which the GC gradient is universal across the eukaryotic *** GC contents at ns sites result in cheaper amino acids,indicating a universal selection for energy efficiency toward the N-termini of encoded *** a genome,the decreasing GC gradient is intensified from lowly to highly expressed genes(more and more protein products),further supporting this *** reveals a conserved selective constraint for cheaper amino acids at the translation start that drives the increased GC contents at ns *** GC contents can facilitate transcription but result in a more stable local secondary structure around the start codon and subsequently impede translation ***,the GC gradients at four-fold and two-fold synonymous sites vary across *** could decrease or increase,suggesting different constraints acting at the GC contents of different codon sites in different *** study reveals that the overall GC contents at the translation start are consequences of complex interactions among several major biological processes that shape the nucleotide sequences,especially efficient energy usage.

关键词： Macroevolution Prioritization of selective forces Energy efficiency Transcription Translation initiation

来源：评论

学校读者我要写书评

暂无评论

Significance-aware Medication Recommendation with Medication Representation Learning

Significance-aware Medication Recommendation with Medication...

引用

International Conference on Computer Supported Cooperative Work in Design

作者： Yishuo Li Zhufeng Shao Weimin Chen Shoujin Wang Yuehan Du Wenpeng Lu Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Mental Health Center Jinan China Data Science Institute University of Technology Sydney Sydney Australia Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China

ISBN: (数字)9798350349184

ISBN: (纸本)9798350349191

The goal of medication recommendation system is to recommend appropriate pharmaceutical interventions based on a patient’s diagnosis. Most of existing approaches often formulate these recommendations use data on diagnoses, procedures, and prescriptions accumulated in the electronic health records (EHR), and despite the great successes, they seem to have limitations on modelling the significance of medication to a patient’s current visit and mining fine-grained medication representation information. To address these issues, we propose a novel Significanceaware Medication Recommendation (SMRec) framework built on significance of medication to patients and fine-grained medication representation learning. Specifically, we first design a encoding mechanism to compute significance information of medications for each patient’s visit. Then, we utilize the set-level medication co-occurrence graph based on patients’ medical history which integrates temporal dependency to learn fine-grained medication representations. Experimental results on the publicly available MIMIC-III dataset demonstrate the superior effectiveness of our model compared to other approaches 1 .

关键词： Representation learning Drugs Federated learning Computational modeling MIMICs Encoding History

来源：评论

学校读者我要写书评

暂无评论

A Semantic-based Replacement for Event Image Privacy 24

A Semantic-based Replacement for Event Image Privacy

引用

24th IEEE International Conference on Computational science and Engineering, CSE 2021

作者： Chen, Zhenfei Tianqing, Zhu Tian, Bing Wang, Yu Ren, Wei China University of Geosciences School of Computer Science Wuhan China Institute of Artificial Intelligence and Blockchain Guangzhou University Guangzhou China China University of Geosciences Guangxi Key Laboratory of Cryptography and Information Security Key Laboratory of Network Assessment Technology CAS School of Computer Science Beijing China

ISBN: (纸本)9781665416603

Thanks to the continuous development of deep learning and the updating of deep neural networks, the accuracy of various computer vision tasks continue improving. On the one hand, the accuracy of image recognition is significantly improved. On the other hand, it also poses a higher challenge of image-based privacy preservation. Although traditional privacy protection methods such as cryptography methods can provide a good privacy protection, they are extremely inconvenient to use and cannot provide good image utility. In order to obtain a balance between image privacy and utility, we propose a privacy-preserving model based on image semantic replacement. We perform semantic replacement or obfuscation to multiple information. Taking the human figures as an example, the information includes faces, scenes, and dressing style. As that information contributes the most to the recognition, we define those items as the privacy of the original image. We replace the event information of the original image, so that the figures in the image can no longer be recognized. With this strategy, the image can still be detected by various detection networks, such as scene detection, which ensures utility. The framework consists of three parts: detection network, scene replacement network, and clothing replacement network. A comprehensive and quantitative experiment set proves the effectiveness of the proposed model. © 2021 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Out-of-Distribution Detection by Principal Component Correspondence

Out-of-Distribution Detection by Principal Component Corresp...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Xiaoyuan Guan Zhiyong Gan Ling Deng Wei Shi Jiankang Chen Shenshen Bu Chunliang Zhao Jianfang Hu Yuren Zhou Wei-Shi Zheng Ruixuan Wang School of Computer Science and Engineering Sun Yat-Sen University Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing MOE Guangzhou China Network Business Group China Unicom Guangzhou China Department of Network Intelligence Pengcheng Laboratory Shenzhen China School of Data and Science Qingdao University of Science and Technology Qingdao China School of Software Engineering Sun Yat-Sen University Zhuhai China

ISBN: (数字)9798350390155

ISBN: (纸本)9798350390162

Out-of-distribution (OOD) detection is vital for the safe application of intelligent systems in real-world scenarios. This paper proposes an enhancement to OOD detection by leveraging the consistency in cognition between two models, both pretrained on in-distribution (ID) data. Specifically, for a given test sample, we first apply Principal Component Analysis (PCA)-based projection on the feature vectors from each model. These obtained feature vectors (with correlation between dimensions decoupled by PCA projection) are then aligned using a multiple linear mapping, which is fitted using the least squares method on the training data. We hypothesize that the regression error for OOD data will be larger than that for ID data, making it a useful metric for OOD detection. Our experimental results demonstrate the effectiveness of this method. When combined with existing robust baselines, our approach achieves state-of-the-art performance in OOD detection.

关键词： Training Measurement Analytical models Training data Vectors data models Intelligent systems Facial features Principal component analysis Least mean squares methods

来源：评论

学校读者我要写书评

暂无评论

An Auto-Parallel Method for Deep Learning Models Based on Genetic Algorithm 29

An Auto-Parallel Method for Deep Learning Models Based on Ge...

引用

29th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2023

作者： Zeng, Yan Huang, Chengchuang Ni, Yijie Zhou, Chunbao Zhang, Jilin Wang, Jue Zhou, Mingyao Xue, Meiting Zhang, Yunquan Hangzhou Dianzi University School of Computer Science and Technology Hangzhou310018 China Ministry of Education Key Laboratory for Modeling and Simulation of Complex Systems Hangzhou310018 China Data Security Governance Zhejiang Engineering Research Center Hangzhou310018 China Hangzhou Dianzi University School of ITMO Joint Institute Hangzhou310018 China Institute of Computer Network Information Center of the Chinese Academy of Sciences Beijing100086 China HuaWei China Institute of Computing Technology of the Chinese Academy of Sciences State Key Laboratory of Computer Architecture Beijing100086 China

ISBN: (纸本)9798350330717

As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or graph algorithms still have issues with search efficiency and applicability. This paper proposes an automatic parallel method based on a dual-population genetic algorithm, TGA, which transforms model partitioning and placement into an integer linear programming problem and constructs a cost model to evaluate the solution. The solution space is built using the neural network's dataflow graph and device cluster's topology, and the dual-population genetic algorithm is used to search for the optimal model parallel strategy. Experiments with various models show that the proposed method can improve single-step execution time by up to 42% compared to the Baechi method and up to 37.7% compared to the Hierarchical method. © 2023 IEEE.

关键词： Auto-Parallel Distributed machine learning Genetic algorithm Integer linear programming Model parallelism

来源：评论

学校读者我要写书评

暂无评论

Disentangled Noisy Correspondence Learning

arXiv

引用

arXiv 2024年

作者： Dang, Zhuohang Luo, Minnan Wang, Jihong Jia, Chengyou Han, Haochen Wan, Herun Dai, Guang Chang, Xiaojun Wang, Jingdong The School of Computer Science and Technology The Ministry of Education Key Laboratory of Intelligent Networks and Network Security The Shaanxi Province Key Laboratory of Big Data Knowledge Engineering Xi'an Jiaotong University Shaanxi Xi'An710049 China SGIT AI Lab State Grid Shaanxi Electric Power Company Limited State Grid Corporation of China Shaanxi China The School of Information Science and Technology University of Science and Technology China United Arab Emirates The Baidu Inc China

Cross-modal retrieval is crucial in understanding latent correspondences across modalities. However, existing methods implicitly assume well-matched training data, which is impractical as real-world data inevitably involves imperfect alignments, i.e., noisy correspondences. Although some works explore similarity-based strategies to address such noise, they suffer from sub-optimal similarity predictions influenced by modality-exclusive information (MEI), e.g., background noise in images and abstract definitions in texts. This issue arises as MEI is not shared across modalities, thus aligning it in training can markedly mislead similarity predictions. Moreover, although intuitive, directly applying previous cross-modal disentanglement methods suffers from limited noise tolerance and disentanglement efficacy. Inspired by the robustness of information bottlenecks against noise, we introduce DisNCL, a novel information-theoretic framework for feature Disentanglement in Noisy Correspondence Learning, to adaptively balance the extraction of MII and MEI with certifiable optimal cross-modal disentanglement efficacy. DisNCL then enhances similarity predictions in modality-invariant subspace, thereby greatly boosting similarity-based alleviation strategy for noisy correspondences. Furthermore, DisNCL introduces soft matching targets to model noisy many-to-many relationships inherent in multi-modal input for noise-robust and accurate cross-modal alignment. Extensive experiments confirm DisNCL's efficacy by 2% average recall improvement. Mutual information estimation and visualization results show that DisNCL learns meaningful MII/MEI subspaces, validating our theoretical analyses. Copyright © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

中国亚热带植被蒸腾驱动力解耦分析

引用

Journal of Geographical sciences 2023年第11期33卷 2159-2174页

作者：金佳鑫蔡裕龙郭熙王龙浩王颖刘元波 College of Hydrology and Water Resources Hohai UniversityNanjing 210024China Key Laboratory of Water Big Data Technology of Ministry of Water Resources Hohai UniversityNanjing 210024China College of Land Resources and Environment Jiangxi Agricultural UniversityNanchang 330045China Institute of Geographic Sciences and Natural Resources Research CASBeijing 100101China School of Tourism and Social Management Nanjing Xiaozhuang UniversityNanjing 211171China Nanjing Institute of Geography and Limnology CASNanjing 210008China National Earth System Science Data Center National Science and Technology Resource Sharing Service PlatformBeijing 100101China

Transpiration(Tc) is a critical component of the global water *** moisture(SM) and vapor pressure deficit(VPD) are key regulators of Tc,and exploring their contributions to changes in Tc can deepen our understanding of the mechanisms of water cycling in terrestrial ***,the driving roles of VPD and SM in Tc changes remain debated because of the coupling of SM and VPD through land-atmosphere interactions which restrict the quantification of the independent effects of SM and VPD on *** decoupling the correlations between SM and VPD using a novel binning approach,this study analyzed the dominant drivers of vegetation transpiration in subtropical China from 2003 to 2018 based on multi-source data,including meteorological reanalysis,remotely sensed soil moisture,transpiration,and land cover *** results show that Tc first increased and then remained stable with an increase in SM across the study area but changed slightly with increasing ***,the relative contribution of SM to the change in Tc was approximately five times that of *** sensitivities of Tc to SM and VPD differed among vegetation *** the sensitivity of Tc to SM was greater than that of VPD for all four vegetation types,the thresholds of Tc in response to SM were different,with the lowest threshold(approximately 35%) for the other forests and the highest threshold(approximately 55%) for short wood *** infer that this is associated with the differences in ecological *** verify the reliability of our conclusions,we used solar-induced chlorophyll fluorescence(SIF) data as a proxy for Tc based on the tight coupling between photosynthesis and *** results were obtained by repeating the *** results of this study,in which the impacts of SM and VPD on Tc were decoupled,are beneficial for further understanding the critical processes involved in water cycling in terrestrial ecosystems in response to climate change.

关键词： transpiration soil moisture vapor pressure deficit decoupling SIF

来源：评论

学校读者我要写书评

暂无评论

Real-Time Semantic Segmentation Algorithm Based on Tversky Loss Function and Mixed Pooling

Real-Time Semantic Segmentation Algorithm Based on Tversky L...

引用

International Conference on networking and network Applications (NaNA)

作者： Peng Xu Ziyi Zhao Sugang Ma Xi'an Science and Technology Museum Xi'an Shaanxi China School of Computer Science and Technology Xi'an University of Posts and Telecommunications Xi'an Shaanxi China Shaanxi Key Laboratory of Network Data Analysis and Intelligent Processing School of Computer Science and Technology Xi'an University of Posts and Telecommunications Xi'an Shaanxi China

In recent years, semantic segmentation methods based on deep learning have made remarkable developments. Despite achieving high segmentation accuracy, the performance of real-time segmentation methods cannot satisfy real-world applications. In order to achieve a balance between segmentation accuracy and speed, a real-time semantic segmentation algorithm based on Tversky loss function and mixed pooling is proposed in this paper. A Short-Term Dense Concatenate network (STDC network) is used to construct the encoder, and a mixed pooling module is used for the final part of the encoder, which uses strip pooling and average pooling to enhance the feature representation while expanding the receptive field. Additionally, a Tversky-based loss function is used for the detail guidance module of the backbone network in the encoder, and a joint loss function is used to supervise the whole network's training. We achieved 76.9% mIoU at 110.7 FPS on the Cityscapes dataset, a 2.4% improvement in accuracy over the benchmark algorithm STDCSeg, and 72.2% mIoU at 177.6 FPS on the Camvid dataset, satisfied the requirements of the real-time segmentation task.

关键词：

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：