检索结果-内蒙古大学图书馆

IEEE International Conference on Healthcare informatics (ICHI)

作者： Xuguang Ai Ramakanth Kavuluru Department of Computer Science University of Kentucky Lexington USA Dept. of Internal Medicine Division of Biomedical Informatics University of Kentucky Lexington USA

End-to-end relation extraction (E2ERE) is an important task in information extraction, more so for biomedicine as scientific literature continues to grow exponentially. E2ERE typically involves identifying entities (or named entity recognition (NER)) and associated relations, while most RE tasks simply assume that the entities are provided upfront and end up performing relation classification. E2ERE is inherently more difficult than RE alone given the potential snowball effect of errors from NER leading to more errors in RE. A complex dataset in biomedical E2ERE is the ChemProt dataset (BioCreative VI, 2017) that identifies relations between chemical compounds and genes/proteins in scientific literature. ChemProt is included in all recent biomedical natural language processing benchmarks including BLUE, BLURB, and BigBio. However, its treatment in these benchmarks and in other separate efforts is typically not end-to-end, with few exceptions. In this effort, we employ a span-based pipeline approach to produce a new state-of-the-art E2ERE performance on the ChemProt dataset, resulting in >4% improvement in F1-score over the prior best effort. Our results indicate that a straightforward fine-grained tokenization scheme helps span-based approaches excel in E2ERE, especially with regards to handling complex named entities. Our error analysis also identifies a few key failure modes in E2ERE for ChemProt.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Comparison of pipeline, sequence-to-sequence, and GPT models for end-to-end relation extraction: experiments with the rare disease use-case

arXiv

引用

arXiv 2023年

作者： Gupta, Shashank Ai, Xuguang Kavuluru, Ramakanth Division of Biomedical Informatics Dept. of Internal Medicine University of Kentucky United States Computer Science Department University of Kentucky United States

Objective: End-to-end relation extraction (E2ERE) is an important and realistic application of natural language processing (NLP) in biomedicine. In this paper, we aim to compare three prevailing paradigms for E2ERE using a complex dataset focused on rare diseases involving discontinuous and nested entities. Methods: We use the RareDis information extraction dataset to evaluate three competing approaches (for E2ERE): NER → RE pipelines, joint sequence to sequence models, and generative pre-trained transformer (GPT) models. We use comparable state-of-the-art models and best practices for each of these approaches and conduct error analyses to assess their failure modes. Results: Our findings reveal that pipeline models are still the best, while sequence-to-sequence models are not far behind;GPT models with eight times as many parameters are worse than even sequence-to-sequence models and lose to pipeline models by over 10 F1 points. Partial matches and discontinuous entities caused many NER errors contributing to lower overall E2E performances. We also verify these findings on a second E2ERE dataset for chemical-protein interactions. Although generative LM-based methods are more suitable for zero-shot settings, when training data is available, our results show that it is better to work with more conventional models trained and tailored for E2ERE. Conclusion: More innovative methods are needed to marry the best of the both worlds from smaller encoder-decoder pipeline models and the larger GPT models to improve E2ERE. As of now, we see that well designed pipeline models offer substantial performance gains at a lower cost and carbon footprint for E2ERE. Our contribution is also the first to conduct E2ERE for the RareDis dataset. The dataset and code for all our experiments are publicly available: https://***/shashank140195/Raredis © 2023, CC BY.

关键词： Carbon footprint

来源：评论

学校读者我要写书评

暂无评论

Integrating Simulation and Optimization: A Case Study in Pamplona for Self-Collection Delivery Points Network Design 35

Integrating Simulation and Optimization: A Case Study in Pam...

引用

35th European Modeling and Simulation Symposium, EMSS 2023

作者： Izco, Irene Serrano-Hernandez, Adrian Sawik, Bartosz Faulin, Javier GILT-OR Group Institute of Smart Cities Dept. Statitistics Computer Science and Mathematics Public University of Navarra Campus Arrosadia s/n Pamplona31006 Spain Department of Business Informatics and Engineering Management AGH University of Science and Technology Mickiewicza 30 Kraków30059 Poland

ISBN: (纸本)9788885741881

The disruptions experienced by the processes in the last mile delivery during the SARS-CoV-2 pandemic raised the dilemma of up-to-date last mile approaches for Urban Logistics (UL) issues. Self-Collection Delivery Systems (SCDS) have been proved to be an improvement for all the players of the SC, providing flexibility of time-windows and reducing overall mileage, delivery time and, consequently, gas emissions. Differing from previous works involving hybrid modeling for automated parcel lockers (APL) network design, this paper brings a System Dynamics Simulation Model (SDSM) to forecast online shopping demand in the Spanish city of Pamplona. A bi-criteria Facility Location Problem (FLP) is solved by means of an e-constraint method, where e is defined as the level of coverage of the total demand. The experiment run considers 90% of demand coverage, in order to obtain the most complex network possible. The simulation and demand forecast was carried out using Anylogic simulation software and the optimization procedure makes use of the Java-based CPLEX API solver. © 2023 The Authors.

关键词： Complex networks

来源：评论

学校读者我要写书评

暂无评论

Leveraging Highly Approximated Multipliers in DNN Inference

arXiv

引用

arXiv 2024年

作者： Zervakis, Georgios Frustaci, Fabio Spantidi, Ourania Anagnostopoulos, Iraklis Amrouch, Hussam Henkel, Jörg The Computer Engineering & Informatics Dept. University of Patras Greece The DIMES University of Calabria Italy The Department of Computer Science Eastern Michigan University United States The School of Electrical Computer and Biomedical Engineering Southern Illinois University Carbondale United States Germany Karlsruhe Institute of Technology Germany

In this work, we present a control variate approximation technique that enables the exploitation of highly approximate multipliers in Deep Neural Network (DNN) accelerators. Our approach does not require retraining and significantly decreases the induced error due to approximate multiplications, improving the overall inference accuracy. As a result, our approach enables satisfying tight accuracy loss constraints while boosting the power savings. Our experimental evaluation, across six different DNNs and several approximate multipliers, demonstrates the versatility of our approach and shows that compared to the accurate design, our control variate approximation achieves the same performance, 45% power reduction, and less than 1% average accuracy loss. Compared to the corresponding approximate designs without using our technique, our approach improves the accuracy by 1.9x on average. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Deep Learning and Optical Character Recognition for Digitization of Meter Reading 12

Deep Learning and Optical Character Recognition for Digitiza...

引用

12th IEEE Symposium on computer Applications and Industrial Electronics, ISCAIE 2022

作者： Chong, Yue Jiet Huat Chua, Kein Babrdel, Mohammad Hau, Lee Cheun Wang, Li Universiti Tunku Abdul Rahman Dept. of Electrical and Electronic Engineering Selangor Malaysia Universiti Tunku Abdul Rahman Dept. of Internet Engineering and Computer Science Selangor Malaysia Universiti Tunku Abdul Rahman Dept. of Mechatronics and BioMedical Engineering Selangor Malaysia National Cheng Kung University Department of Electrical Engineering Tainan Taiwan

ISBN: (纸本)9781665487030

One of the important steps in digital transformation is to make all the instrumental devices connected to the cloud. However, replacing the existing analogue meters with the cloud-connected digital meters can be very costly especially for industrial grade meters. In this project, a deep learning model based on Single Shot Detector (SSD) MobileNet V2 and an optical character recognition (Tesseract OCR) engine are developed for the low-cost digitization of analogue meter readings. The deep learning model is trained with a dataset of 750 meters' images, and it is used to detect the region of interest where the meter's readings are located. The OCR is used to convert the readings to string datatype. The image processing techniques via the OpenCV library are adopted for enhancing the quality of the region of interest (ROI). The model is developed in Python and the evaluation is carried for various types of meters, illumination conditions, and backgrounds. The results show that the deep learning model and OCR accuracies are 95% and 93%, respectively. © 2022 IEEE.

关键词： Optical character recognition

来源：评论

学校读者我要写书评

暂无评论

A Study on Neuro-Symbolic Artificial Intelligence: Healthcare Perspectives

arXiv

引用

arXiv 2025年

作者： Hossain, Delower Chen, Jake Y. Dept. of Computer Science The University of Alabama Birmingham United States Dept. of Genetic Biomedical Engineering Informatics and Data Science Heersink School of Medicine The University of Alabama Birmingham United States Systems Pharmacology AI Research Center The University of Alabama Birmingham United States

Over the last few decades, Artificial Intelligence (AI) scientists have been conducting investigations to attain human-level performance by a machine in accomplishing a cognitive task. Within machine learning, the ultimate aspiration is to attain Artificial General Intelligence (AGI) through a machine. This pursuit has led to the exploration of two distinct AI paradigms. Symbolic AI, also known as classical or GOFAI (Good Old-Fashioned AI) and Connectionist (Sub-symbolic) AI, represented by Neural Systems, are two mutually exclusive paradigms. Symbolic AI excels in reasoning, explainability, and knowledge representation but faces challenges in processing complex real-world data with noise. Conversely, deep learning (Black-Box systems) research breakthroughs in neural networks are notable, yet they lack reasoning and interpretability. Neuro-symbolic AI (NeSy), an emerging area of AI research, attempts to bridge this gap by integrating logical reasoning into neural networks, enabling them to learn and reason with symbolic representations. While a long path, this strategy has made significant progress towards achieving common sense reasoning by systems. This article conducts an extensive review of over 977 studies from prominent scientific databases (DBLP, ACL, IEEExplore, Scopus, PubMed, ICML, ICLR), thoroughly examining the multifaceted capabilities of Neuro-Symbolic AI, with a particular focus on its healthcare applications, particularly in drug discovery, and Protein engineering research. The survey addresses vital themes, including reasoning, explainability, integration strategies, 41 healthcare-related use cases, benchmarking, datasets, current approach limitations from both healthcare and broader perspectives, and proposed novel approaches for future experiments. Moreover, it identifies critical challenges in applying Neuro-Symbolic AI to medicine and represents a comprehensive exploration of its transformative potential in the biomedical field. © 2025, CC BY-NC

关键词： Knowledge representation

来源：评论

学校读者我要写书评

暂无评论

Product Recommendation Using Collaborative Filtering and K-Means Clustering

Product Recommendation Using Collaborative Filtering and K-M...

引用

Computing, Power and Communication Technologies (IC2PCT), IEEE International Conference on

作者： M Ranjith Kumar S Vishnu G Roshen D Niranjan Kumar P Revathi David Ricardo Ledo Baster Department of Mathematics Amrita School of Engineering Amrita Vishwa Vidyapeetham Chennai India Dept. of Computer Science & Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Chennai India Dept. of Computer Science & Engineering C. Abdul Hakeem College of Engineering and Technology Melvisharam Vellore India Dept. of Informatics and Mathematics University of Holguin Cuba

The growth of e-commerce has altered how consumers shop, providing a digital space where convenience, vast product offerings, and competitive pricing converge. In today’s world, e-commerce websites are transitioning from traditional search-driven methods to customized and intuitive approaches via product suggestions. Product recommendation systems are vital in e-commerce, from bringing new business to retaining existing ones. Our three-part recommendation system is designed so that new users have a great and engaging experience as the Product Popularity -Based System shows them carefully chosen products that are in demand. Collaborative Filtering Recommendations are highly personalized recommendations given to people who have already made their first purchases based on their prior actions and preferences. The K-Means Clustering-Based Recommendation System uses textual clustering analysis to deliver contextually relevant recommendations. We use a variety of evaluation metrics, such as click-through rates, user engagement, and the Silhouette Score, to assess the effectiveness and accuracy of our recommendation systems. Our findings show significant increases in user engagement, conversion rates, and relevant recommendations. Our findings demonstrate the transformative power of well-designed recommendation systems, which improve user experiences and retention and provide invaluable solutions for businesses entering the e-commerce space. This paper provides an in-dept. examination of the multifaceted landscape of e-commerce recommendations, shedding light on their far-reaching implications for customer acquisition and retention in this dynamic digital era.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Design of efficient single-etch grating couplers for silicon nitride photonics at 1550 nm

Design of efficient single-etch grating couplers for silicon...

引用

2024 Photonics North, PN 2024

作者： Fraser, William Benedikovic, Daniel Korcek, Radovan Milanizadeh, Maziyar Bucio, Thalía Domínguez Vitali, Valerio Gardes, Frederic Y. Schmid, Jens H. Cheben, Pavel Ye, Winnie N. Dept. Electronics Carleton University Ottawa Canada Dept. Multimedia and Information-Communication Technologies University of Zilina Zilina Slovakia University Science Park University of Zilina Zilina Slovakia National Research Council Ottawa Canada Optoelectronics Research Centre University of Southampton Highfield Campus Southampton United Kingdom Electrical Computer and Biomedical Engineering Department University of Pavia Pavia27100 Italy

ISBN: (纸本)9798350366365

We present the design of single-etch grating couplers implemented on native and a hybrid silicon nitride (SiN) photonic platform with amorphous silicon (α-Si) overlays, operating in the telecom waveband. The SiN-only grating coupler has a coupling loss of -3.6 dB and a 1-dB bandwidth of 37 nm. A hybrid α-Si-on-SiN platform yields improved coupling loss of -1.4 dB and -1 dB for uniform and subwavelength grating apodized designs, respectively. © 2024 IEEE.

关键词： Silicon nitride

来源：评论

学校读者我要写书评

暂无评论

Instance-Level Safety-Aware Fidelity of Synthetic Data and its Calibration

Instance-Level Safety-Aware Fidelity of Synthetic Data and i...

引用

International Conference on Intelligent Transportation

作者： Chih-Hong Cheng Paul Stöckel Xingyu Zhao Dept. of Computer Science & Engineering Chalmers & Univ. of Gothenburg Sweden Department of Informatics University of Hildesheim Germany WMG University of Warwick United Kingdom

ISBN: (数字)9798331505929

ISBN: (纸本)9798331505936

Modeling and calibrating the fidelity of synthetic data is paramount in shaping the future of safe and reliable self-driving technology by offering a cost-effective and scalable alternative to real-world data collection. We focus on its role in safety-critical applications, introducing four types of instance-level fidelity that go beyond mere visual input characteristics. The aim is to ensure that applying testing on synthetic data can reveal real-world safety issues, and the absence of safety-critical issues when testing under synthetic data can provide a strong safety guarantee in real-world behavior. We suggest an optimization method to refine the synthetic data generator, reducing fidelity gaps identified by deep learning components. Experiments show this tuning enhances the correlation between safety-critical errors in synthetic and real data.

关键词： Visualization Correlation Semantics Optimization methods Data collection Generators Safety Reliability Synthetic data Testing

来源：评论

学校读者我要写书评

暂无评论

Using annotation for computerized support for fast skimming of cardiology electronic health record notes

Using annotation for computerized support for fast skimming ...

引用

2023 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2023

作者： Dehkordi, Mahshad Koohi H. Einstein, Andrew J. Zhou, Shuxin Elhanan, Gai Perl, Yehoshua Keloth, Vipina K. Geller, James Liu, Hao Njit Department of Computer Science NewarkNJ United States Irving Medical Center Cardiology Division Dept. of Medicine New YorkNY United States Desert Research Institute Renown Institute for Health Innovation RenoNV United States Yale School of Medicine Biomedical Informatics and Data Science New HavenCT United States Njit Department of Data Science NewarkNJ United States Montclair State University Department of Computer Science MontclairNJ United States

ISBN: (纸本)9798350337488

Under the circumstances prevalent in current healthcare, medical professionals such as physicians and nurses need to read large numbers of Electronic Health Record (EHR) notes. This demand fosters a situation in which providers typically do not read a whole note but quickly skim it, to capture its essential content. Consequently, by fast skimming, one may miss a critical medical fact. Annotation highlights important content in EHR notes, and enables healthcare professionals to perform fast skimming, thereby minimizing the risk of missing critical information, which is detrimental to patient care. We designed the Cardiology Interface Terminology (CIT) for the purpose of annotation of cardiology EHRs. We emphasize that by annotation we refer to highlighting important information in the EHR notes for enabling fast skimming rather than just recognizing names of diseases, drugs, etc. from Reference Terminologies as is usually done by existing Named Entity Recognition (NER) systems. The CIT design starts with the cardiology components of SNOMED CT. It is enhanced by mining phrases from cardiology EHRs, as potential CIT concepts, which are of higher granularity than SNOMED concepts. Machine learning (ML), the state of the art technique for mining concepts from EHRs, requires training data. However, there is no training data for designing CIT. In the first stage, we introduce an innovative semi-automatic method for mining concepts from EHRs, to replace costly manual mining. The only manual portion is the review of the automatically mined phrases, before their insertion as CIT concepts. The effectiveness of annotation of cardiology EHRs with CIT was evaluated utilizing proper metrics, and compared to annotation with SNOMED CT. In a future second stage, ML mining techniques will be used for enhancing CIT with extra concepts from EHRs, utilizing the concepts added in the first stage as training data. This work focuses on a novel semi-automated method to design the Cardiology I

关键词： annotation cardiology EHR fast skimming

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：