检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Sukhija, Bhavya Turchetta, Matteo Lindner, David Krause, Andreas Trimpe, Sebastian Baumann, Dominik Department of Computer Science ETH Zürich Switzerland Institute for Data Science in Mechanical Engineering RWTH Aachen University Germany Department of Electrical Engineering and Automation Aalto University Espoo Finland Department of Information Technology Uppsala University Sweden

Learning optimal control policies directly on physical systems is challenging. Even a single failure can lead to costly hardware damage. Most existing model-free learning methods that guarantee safety, i.e., no failures, during exploration are limited to local optima. This work proposes GoSafeOpt as the first provably safe and optimal algorithm that can safely discover globally optimal policies for systems with high-dimensional state space. We demonstrate the superiority of GoSafeOpt over competing model-free safe learning methods in simulation and hardware experiments on a robot arm. © 2022, CC BY-NC-ND.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Prompt Tuning Pushes Farther, Contrastive Learning Pulls Closer: A Two-Stage Approach to Mitigate Social Biases

arXiv

引用

arXiv 2023年

作者： Li, Yingji Du, Mengnan Wang, Xin Wang, Ying College of Computer Science and Technology Jilin University Changchun China Department of Data Science New Jersey Institute of Technology Newark United States School of Artificial Intelligence Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun China

As the representation capability of Pre-trained Language Models (PLMs) improve, there is growing concern that they will inherit social biases from unprocessed corpora. Most previous debiasing techniques used Counterfactual data Augmentation (CDA) to balance the training corpus. However, CDA slightly modifies the original corpus, limiting the representation distance between different demographic groups to a narrow range. As a result, the debiasing model easily fits the differences between counterfactual pairs, which affects its debiasing performance with limited text resources. In this paper, we propose an adversarial training-inspired two-stage debiasing model using Contrastive learning with Continuous Prompt Augmentation (named CCPA) to mitigate social biases in PLMs’ encoding. In the first stage, we propose a data augmentation method based on continuous prompt tuning to push farther the representation distance between sample pairs along different demographic groups. In the second stage, we utilize contrastive learning to pull closer the representation distance between the augmented sample pairs and then fine-tune PLMs’ parameters to get debiased encoding. Our approach guides the model to achieve stronger debiasing performance by adding difficulty to the training process. Extensive experiments show that CCPA outperforms baselines in terms of debiasing performance. Meanwhile, experimental results on the GLUE benchmark show that CCPA retains the language modeling capability of PLMs. Copyright © 2023, The Authors. All rights reserved.

关键词： Encoding (symbols)

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Pretraining for Biomedical Term Embeddings

arXiv

引用

arXiv 2023年

作者： Cai, Bryan Zeng, Sihang Lin, Yucong Yuan, Zheng Zhou, Doudou Tian, Lu Department of Computer Science Stanford University United States Department of Electronic Engineering Tsinghua University Beijing China Institute of Engineering Medicine Beijing Institute of Technology Beijing China Alibaba Damo Academy Hangzhou China Department of Biostatistics Harvard T.H. Chan School of Public Health United States Department of Biomedical Data Science Stanford University United States

Electronic health records (EHR) contain narrative notes that provide extensive details on the medical condition and management of patients. Natural language processing (NLP) of clinical notes can use observed frequencies of clinical terms as predictive features for downstream applications such as clinical decision making and patient trajectory prediction. However, due to the vast number of highly similar and related clinical concepts, a more effective modeling strategy is to represent clinical terms as semantic embeddings via representation learning and use the low dimensional embeddings as feature vectors for predictive modeling. To achieve efficient representation, fine-tuning pretrained language models with biomedical knowledge graphs may generate better embeddings for biomedical terms than those from standard language models alone. These embeddings can effectively discriminate synonymous pairs of from those that are unrelated. However, they often fail to capture different degrees of similarity or relatedness for concepts that are hierarchical in nature. To overcome this limitation, we propose HIPRBERT, a novel biomedical term representation model trained on additionally complied data that contains hierarchical structures for various biomedical terms. We modify an existing contrastive loss function to extract information from these hierarchies. Our numerical experiments demonstrate that HIPRBERT effectively learns the pair-wise distance from hierarchical information, resulting in a substantially more informative embeddings for further biomedical applications. Copyright © 2023, The Authors. All rights reserved.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

DISC-LawLLM: Fine-tuning Large Language Models for Intelligent Legal Services

arXiv

引用

arXiv 2023年

作者： Yue, Shengbin Chen, Wei Wang, Siyuan Li, Bingxuan Shen, Chenchen Liu, Shujun Zhou, Yuxuan Xiao, Yao Yun, Song Huang, Xuanjing Wei, Zhongyu Research Institute of Intelligent Complex Systems Fudan University China Shanghai Center for Mathematical Sciences Fudan University China School of Software Engineering Huazhong University of Science and Technology China School of Data Science Fudan University China School of Computer Science Fudan University China Rule of Law Institute Northwest University of Political and Law China New York University Shanghai China

We propose DISC-LawLLM, an intelligent legal system utilizing large language models (LLMs) to provide a wide range of legal services. We adopt legal syllogism prompting strategies to construct supervised fine-tuning datasets in the Chinese Judicial domain and fine-tune LLMs with legal reasoning capability. We augment LLMs with a retrieval module to enhance models' ability to access and utilize external legal knowledge. A comprehensive legal benchmark, DISC-Law-Eval, is presented to evaluate intelligent legal systems from both objective and subjective dimensions. Quantitative and qualitative results on DISC-Law-Eval demonstrate the effectiveness of our system in serving various users across diverse legal scenarios. The detailed resources are available at https://***/FudanDISC/DISC-LawLLM. Copyright © 2023, The Authors. All rights reserved.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Dynamic Lane Management with IoT for Real-time Lane Configuration and Traffic Flow

Dynamic Lane Management with IoT for Real-time Lane Configur...

引用

computer, Communication and Control (IC4), International Conference on

作者： Mohammed Rehaman Pasha S. Latha R. Usha V. Bini Marin M.V Ishwarya M. Muthulekshmi Department of Computer Science and Engineering Malla Reddy Institute of Engineering and Technology Hyderabad Telangana India Department of Department of Biotechnology Vinayaka Mission’s Kirupananda Variyar Engineering College Vinayaka Mission’s Research Foundation(DU) Salem Tamil Nadu India Department of Mathematics S. A. Engineering College Chennai Tamil Nadu India Faculty of Management Studies SRM Institute of Science and Technology Ramapuram Chennai Tamil Nadu India Artificial intelligence and Data science Department Agni College of Technology Thazhambur Tamil Nadu India Department of Biomedical Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Chennai Tamil Nadu India

Traffic may be reduced and transportation efficiency improved by properly controlling traffic flow in urban areas. This research provides a new method for managing lanes in real-time by using IoT connectivity. To optimize traffic flow in response to shifting demand patterns, the suggested system prioritizes making lane design changes in real time depending on the current traffic conditions. Using information collected from Internet of Things (IoT) sensors and linked automobiles, the system may dynamically allot lanes in response to variations in traffic volumes and congestion. It describes developing and deploying a dynamic lane control system based on the Internet of Things, stressing the technology’s potential to improve traffic flow in congested metropolitan areas. The simulation results show its efficiency in decreasing traffic and increasing throughput. It demonstrates the potential of intelligent transportation systems for creating smarter and more efficient cities by combining IoT, real-time data processing, and traffic control.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Construction Equipment Performance with Cloud data Analysis for Predictive Maintenance

Construction Equipment Performance with Cloud Data Analysis ...

引用

IEEE Recent Advances in Intelligent Computational Systems (RAICS)

作者： Senthil Kumar Seeni P. Maheswari Jothibabu Konidhala M. Raja Suguna M.V. Ishwarya S. Srinivasan Department of Mobile Application Development Cognizant Technology Solutions Buffalo Grove Illinois USA Faculty of Management Studies SRM Institute of Science and Technology Kattankulathur Chennai India Department of Electronics and Communication Engineering University of Technology and Applied Sciences- Al Musannah Musanna South Al Batina Oman Department of Computer Science and Engineering RMK College of Engineering and Technology Chennai Tamil Nadu India Artificial intelligence and Data science Department Agni College of Technology Chennai Tamil Nadu India Department of Biomedical Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Chennai Tamil Nadu India

ISBN: (数字)9798350381689

ISBN: (纸本)9798350381696

The research uses Internet of Things technology for predictive maintenance in the construction sector. The system addresses incorporating Internet of Things sensors into construction machinery to enable real-time monitoring of operating parameters. The study suggests a proactive maintenance strategy that may anticipate equipment breakdowns by examining the data obtained and using sophisticated prediction models. With this strategy, construction teams can arrange maintenance activities systematically, which cuts down on downtime and helps avoid expensive interruptions. The abstract emphasizes the advantages of this practice, which include reducing unexpected downtime as much as possible, extending the life of equipment, and making the most of available resources. The study proves the usefulness of the IoT-based predictive maintenance approach via empirical investigations and the analysis of case studies. This validates the strategy's efficacy, contributing to enhanced equipment performance and increased work efficiency in the construction industry.

关键词： Electric breakdown Predictive models Real-time systems Sensor systems Maintenance Sensors Trajectory Internet of Things Machinery Predictive maintenance

来源：评论

学校读者我要写书评

暂无评论

Evaluation of Cloud Recovery System for Financial Applications Based on Hadoop

Evaluation of Cloud Recovery System for Financial Applicatio...

引用

IEEE International Symposium on Telecommunication Technologies (ISTT)

作者： LU Li LU Tong Aduwati Sali School of Microelectronics and Data Science Anhui University of Technology Maanshan China Department of Computer and Communication Systems Engineering Faculty of Engineering Wireless and Photonics Networks Research Centre of Excellence (WiPNET) Universiti Putra Malaysia Serdang Selangor Malaysia Institute for Mathematical Research (INSPEM) Universiti Putra Malaysia Serdang Selangor Malaysia

ISBN: (数字)9798350392074

ISBN: (纸本)9798350392081

The rise of cloud-based disaster recovery systems has reshaped disaster recovery practices. Integrated into the Hadoop framework with core components like Hadoop Distributed File System (HDFS), MapReduce, and Yet Another Resource Negotiator (YARN), these systems ensure data integrity and business continuity during crises while enhancing data reliability and operational efficiency, surpassing traditional recovery systems. This study presents a comprehensive comparative analysis of traditional and Hadoop-integrated cloud recovery systems, focusing on hardware architecture and performance. Architecturally, it highlights the stark differences between conventional models and Hadoop-based innovations. The results show that the Hadoop-backed recovery system exhibits remarkable improvement, reducing Recovery Time Objective (RTO) from over 18 seconds to under 7 seconds while preserving data integrity as per Recovery Point Objective (RPO) standards. This finding underscores the academic importance of evaluating cloud-based disaster recovery within the Hadoop context, offering insights into its transformative potential for disaster recovery strategies.

关键词： Technological innovation Disasters data integrity Switches Attenuation Hardware Business continuity Yarn Millimeter wave communication Standards

来源：评论

学校读者我要写书评

暂无评论

Incremental Reduction Methods Based on Granular Ball Neighborhood Rough Sets and Attribute Grouping

SSRN

引用

SSRN 2023年

作者： Li, Yan Wu, Xiaoxue Wang, Xizhao Research Center for Applied Mathematics and Interdisciplinary Sciences Beijing Normal University Zhuhai Campus Zhuhai519087 China College of Mathematics and Information Science Hebei University Baoding071002 China Big Data Institute College of Computer Science and Software Engineering Shenzhen University Guangdong Shenzhen518060 China

In the framework of rough sets, incremental algorithms can effectively reduce repetitive computations for dynamic datasets by discovering the update principles of relevant *** the variations in attribute sets, incremental methods based on neighborhood granulation provide fast updates to attribute reduction in continuous-valued information systems. However, most of these methods assume that the sample distribution and neighborhood radius remain fixed with changes in the attribute set, which may mislead the granulation process and affect the model’s classification ability. In this paper, a new incremental reduction method based on granular balls and attribute grouping is proposed for dynamic information systems with multiple attribute additions. Different neighborhood radii are adaptively determined when the attribute set changes, and the number of neighborhood granules can also be effectively ***, the attributes are grouped based on the k-means algorithm, and only attributes from different groups or with small relevance to those in the current reduction set are considered to be incorporated as a reduction attribute, thus reducing the computation time and simultaneously preserving informative *** and nonlinear correlation coefficients are used to measure attribute relevance based on which three corresponding incremental reduction algorithms are ***, extensive experimental results on 12 benchmark datasets are shown to compare the proposed method with non-incremental and typical incremental methods in terms of time cost, classification accuracy, the size of attribute reduction, and coverage of the reduction. © 2023, The Authors. All rights reserved.

关键词： Granulation

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-based Identification and Classification of Rice Plant Diseases for Precision Agriculture

Deep Learning-based Identification and Classification of Ric...

引用

Automation, Computing and Renewable Systems (ICACRS), International Conference on

作者： N. Venkatesh Swagata Sarkar A Sunil T.Guru Priya R Aarthi K.Mekala Devi School of CS&AI SR University &Warangal Artificial intelligence and Data Science Sri Sairam Engineering College Sai Leo Nagar West Tambaram Chennai computer science and engineering rajarajeswari college of engineering Ramohalli cross Bangalore Mysore road Kumbalgod Karnataka Electronics and Communication Engineering Periyar Maniammai Institute of Science and Technology PMIST VALLAM Department of ECE R.M.D Engineering College Chennai ECE Department S.A. Engineering college Chennai

Because they reduce both the quality and yield of crops, plant diseases seriously threaten our food supply’s reliability. Plant diseases often result in a significant reduction or complete absence of grain yield in various agricultural contexts. Classifying rice plant diseases and insects pose a significant challenge due to the intricate structure of their composition and the striking resemblance 5bserved among various disease types and insect species. Diseases and insects in crops must be discovered and categorized as soon as possible, ideally before they may spread. There are a number of methods for spotting plant diseases, but deep learning is now the gold standard. The authors provide a novel deep-learning approach for detecting diseases in rice fields using photographs. Convolutional Neural Networks with attention mechanism (CNNAM) can be used to detect diseases in rice plants. Due to its reliability in picture identification and classification, CNNs have been widely employed by researchers for plant disease identification. The identification of rice plant diseases, however, has only been the subject of a small number of studies. This research illuminates the present state of rice plant disease and the methods of disease detection based on Deep Learning. Deep learning models excel about more conventional machine learning methods. With the suggested model, authors were able to categorize rice photos with a 99.8 percent accuracy by learning latent patterns in the raw images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficient Feature Extraction for Classification of Alzheimer's Abnormalities Using Deep Learning

Efficient Feature Extraction for Classification of Alzheimer...

引用

Emerging Technology (INCET), International Conference for

作者： Minal A. Zope Rakesh K. Deshmukh Priya Pise Shrikant D. Bhopale Dept. of Computer Engineering AISSMS's Institute of Information Technology Kalinga University Raipur Pune India Kalinga University Raipur India Artificial Intelligence and Data Science Indira College of Engineering and Management Pune India Dr. D. Y. Patil Pratisthan's Colleg of Engineering Salokhenagar Kolhapur India

ISBN: (数字)9798350361155

ISBN: (纸本)9798350361162

Recent machine learning AD predictions employed feature selection and extraction methods. The literature research revealed several attempts to identify AD by using different Deep learning layer and micro-simulation techniques. However, identifying important features that can detect AD extremely early remains a difficult task. Automation, poor classification accuracy, over fitting, etc. are the other challenges while using machine learning algorithms. The recent progression of using deep learning methods has received the significant attention for the automatic AD classification. CNN's strategy, while successful, has certain drawbacks: (1) The CNN model uses local receptive fields, shares weights, and subsamples, (2) deep neural networks need a lot of labeled sample training models to avoid over fitting, which may cause the loss of crucial information like representative features at various frequencies and orientations; and (3) the higher computational overhead is another challenge. Therefore, these challenges motivate us the need of the proposed study. The scope of the proposed study is limited to the classification of AD from the input medical images.

关键词： Deep learning Training Machine learning algorithms Automation Computational modeling Fitting Artificial 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：