检索结果-内蒙古大学图书馆

1st Workshop on Large Foundation Models for Educational Assessment

作者： Gao, Rujun Guo, Xiaosu Li, Xiaodi Narayanan, Arun Balajiee Lekshmi Thomas, Naveen Srinivasa, Arun R. J. Mike Walker’66 Department of Mechanical Engineering Texas A&M University United States Computer Science Department University of Texas Dallas United States Intelligent Systems Program University of Pittsburgh United States Department of Artificial Intelligence and Informatics Mayo Clinic United States

This study explores the feasibility of using large language models (LLMs), specifically GPT-4o (ChatGPT), for automated grading of conceptual questions in an undergraduate Mechanical Engineering course. We compared the grading performance of GPT-4o with that of human teaching assistants (TAs) on ten quiz problems from the MEEN 361 course at Texas A&M University, each answered by approximately 225 students. Both the LLM and TAs followed the same instructor-provided rubric to ensure grading consistency. We evaluated performance using Spearman’s rank correlation coefficient and Root Mean Square Error (RMSE) to assess the alignment between rankings and the accuracy of scores assigned by GPT-4o and TAs under zero- and few-shot grading settings. In the zero-shot setting, GPT-4o demonstrated a strong correlation with TA grading, with Spearman’s rank correlation coefficient exceeding 0.6 in seven out of ten datasets and reaching a high of 0.9387. Our analysis reveals that GPT-4o performs well when grading criteria are straightforward but struggles with nuanced answers, particularly those involving synonyms not present in the rubric. The model also tends to grade more stringently in ambiguous cases compared to human TAs. Overall, ChatGPT shows promise as a tool for grading conceptual questions, offering scalability and consistency. © 2024 R. Gao1, X. Guo2, X. Li4, A.B.L. Narayanan3, N. Thomas1 & A.R. Srinivasa1.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

The Banking Industry is Undergoing a Digital Transformation as Part of Industrial 4.0 10

The Banking Industry is Undergoing a Digital Transformation ...

引用

10th International Conference on Smart Computing and Communication, ICSCC 2024

作者： Sardjono, Wahyu Pranata, Dewi Sagita Ramdhana, Aditya Agung Dwi BINUS Graduate Program - Master of Information Systems Management Bina Nusantara University Information Systems Management Department Jakarta11480 Indonesia Mathematical & Engineering Sciences King's College London Cyber Security Studi Program Faculty of Natural Department of Informatics West Minster Strands WC2R 2LS United Kingdom University of Airlangga Master of Management Faculty of Economics and Business Surabaya60286 Indonesia

ISBN: (纸本)9798350363104

In the current day, digital transformation is imperative;both customer demand and market competitiveness are mostly driving that trend. The fourth industrial revolution has also affected the financial sector. One difficulty that has to be addressed is the desire from customers for quick and useful services. Traditional procedures need to be replaced with digital ones. Furthermore, banking organizations will have a competitive edge in the current market if they are successful in this digital transformation process. Successful transformation, however, does not happen by accident;in fact, the likelihood of completing this change is seen to be very low, necessitating rigorous planning for a company that is ready for the future. There are several components to the preparation in question, both internal and external. In this situation, the company's ability to combine and rearrange its resources quickly is essential for adapting to the ever-changing business environment. © 2024 IEEE.

关键词： Commerce

来源：评论

学校读者我要写书评

暂无评论

Feature Engineering for Interpretable Machine Learning for Quality Assurance in Radiation Oncology 18

Feature Engineering for Interpretable Machine Learning for Q...

引用

18th World Congress on Medical and Health informatics: One World, One Health - Global Partnership for Digital Innovation, MEDINFO 2021

作者： Pillai, Malvika Adapa, Karthik Shumway, John W. Dooley, John Das, Shiva K. Chera, Bhishamjit S. Mazur, Lukasz Carolina Health Informatics Program University of North Carolina Chapel HillNC United States Department of Radiation Oncology University of North Carolina Chapel HillNC United States

ISBN: (纸本)9781643682648

Chart checking is a time intensive process with high cognitive workload for physicists. Previous studies have partially automated and standardized chart checking, but limited studies implement data-driven approaches to reduce cognitive workload for quality assurance processes. This study aims to evaluate feature selection methods to improve the interpretability and transparency of machine learning models in predicting the degree of difficulty for a pretreatment physics chart check. We compare chi-square, mutual information, feature importance thresholding, and greedy feature selection for four different classifiers. Random forest has the highest performance with SMOTE oversampling using mutual information for feature selection (accuracy 84.0%, AUC 87.0%, precision 80.0%, recall 80.0%). This study demonstrates that feature selection methods can improve model interpretability and transparency. © 2022 International Medical informatics Association (IMIA) and IOS Press.

关键词： Quality assurance

来源：评论

学校读者我要写书评

暂无评论

Evaluating an e-Learning Platform at Graduate School Based on User Experience Evaluation Technique 3

Evaluating an e-Learning Platform at Graduate School Based o...

引用

3rd International Conference on Electrical, Electronics, informatics, and Vocational Education, ICE-ELINVO 2020

作者： Abubakari, M.S. Nurkhamid Hungilo, G. Department of Electronics Engineering Education Postgraduate Program Universitas Negeri Yogyakarta Indonesia Department of Informatics Engineering Graduate Program University Atma Jaya Yogyakarta Indonesia

This study was aimed to evaluate the e-learning platform in a graduate school by using User Experience Questionnaire (UEQ). The data were collected from 65 respondents and analyzed using an excel data analysis tool. The results indicate that all six scales of the instrument are consistent except the novelty scale. Also, the calculated means of the scales are slightly positive as all values are greater than 0.8 compared to the UEQ benchmark. Although the respondents show a little positive impression with the platform, all scales means are below average category except efficiency scale which is above average as compared to the benchmark. This indicates that the platform needs more improvement in all aspects of user experience for it to be at least in a good category. The findings discussed in this paper are valuable for all academic stakeholders in various educational institutions which implement interactive e-learning systems as well as practitioners of user experience and developers of e-learning systems. © 2021 Published under licence by IOP Publishing Ltd.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

U-Netmer: U-Net Meets Transformer for Medical Image Segmentation

SSRN

引用

SSRN 2024年

作者： He, Sheng Bo, Rina Grant, P. Ellen Ou, Yangming Boston Children’s Hospital Harvard Medical School 300 Longwood Ave. BostonMA United States Computational Health Informatics Program

The combination of the U-Net based deep learning models and Transformer is a new trend for medical image segmentation. U-Net can extract the detailed local semantic and texture information and Transformer can learn the long-rang dependencies among pixels in the input image. However, directly adapting the Transformer for segmentation has "token-flatten" problem (flattens the local patches into 1D tokens which losses the interaction among pixels within local patches) and "scale-sensitivity" problem (uses a fixed scale to split the input image into local patches). Compared to directly combining U-Net and Transformer, we propose a new global-local fashion combination of U-Net and Transformer, named U-Netmer, to solve the two problems. The proposed U-Netmer splits an input image into local patches. The global-context information among local patches is learnt by the self-attention mechanism in Transformer and U- Net segments each local patch instead of flattening into tokens to solve the ‘token-flatten" problem. The U-Netmer can segment the input image with different patch sizes with the identical structure and the same parameter. Thus, the U-Netmer can be trained with different patch sizes to solve the "scale-sensitivity" problem. We conduct extensive experiments in 7 public datasets on 7 organs (brain, heart, breast, lung, polyp, pancreas and prostate) and 4 imaging modalities (MRI, CT, ultrasound, and endoscopy) to show that the proposed U-Netmer can be generally applied to improve accuracy of medical image segmentation. These experimental results show that U-Netmer provides state-of-the-art performance compared to baselines and other models. In addition, the discrepancy among the outputs of U-Netmer with different scales is linearly correlated to the segmentation accuracy which can be considered as a confidence score to rank test images by difficulty without ground-truth. The code will be available on GitHub. © 2024, The Authors. All rights reserved.

关键词： Computerized tomography

来源：评论

学校读者我要写书评

暂无评论

Method for improvement of transparency: Use of text mining techniques for reclassification of governmental expenditures records in Brazil

引用

International Journal of Business Intelligence and Data Mining 2021年第2期18卷 155-196页

作者： de Oliveira Almeida, Gustavo Revoredo, Kate Cappelli, Claudia Maciel, Cristiano Graduate Program in Business Federal Fluminense University Niteroi/RJ Brazil Graduate Program in Informatics Federal University of Rio de Janeiro State Rio de Janeiro/RJ Brazil Computing Institute Mato Grosso Federal University Cuiabá/MT Brazil

Many countries have transparency laws requiring availability of data. However, often data is available but not transparent. We present the Transparency Portal of Brazilian Federal Government case and discuss limitations of public acquisitions data stored in free text format. We employed text-mining techniques to reclassify descriptive texts of measurement units related to products and services. The solution presented in KNIME and Java aggregated measurements in the original (n = 69,372 with 78% reduction in number of descriptions, 94% items classified) and in cross validation sample (n = 105,266 with 88% reduction, classifying 78% of items). In addition, we tested computational time for processing of texts for a wide range of data input sizes, suggesting the stability and scalability of the solution to process larger datasets. Finally, we produced analysis identifying probable input errors, suppliers and purchasing units with abnormal transactions and factors affecting procurement prices. We present suggestions for future research and improvements. Copyright © 2021 Inderscience Enterprises Ltd.

关键词： Transparency

来源：评论

学校读者我要写书评

暂无评论

Multi Task Classification Using Deep Learning Approaches for Big Five Personality Traits Prediction: A Review 11

Multi Task Classification Using Deep Learning Approaches for...

引用

11th International Conference on Information Technology, Computer and Electrical Engineering, ICITACEE 2024

作者： Nudin, Salamun Rohman Gernowo, Rahmat Somantri, Maman Wibowo, Adi Doctoral Program of Information System School of Postgraduate Studies Diponegoro University Semarang Indonesia Diponegoro University Faculty of Science and Mathematics Department of Physics Semarang Indonesia Diponegoro University Faculty of Engineering Department of Electrical Engineering Semarang Indonesia Diponegoro University Faculty of Science and Mathematics Department of Informatics Semarang Indonesia

ISBN: (纸本)9798350389289

This paper presents a comprehensive review of deep learning methods for the Big Five personality traits prediction using multi-task classification. The purpose of the review is to determine the performance of models developed in previous research to describe research opportunities and challenges in the future. The Big Five model is a framework in psychology that categorizes human personality into five dimensions: openness, conscientiousness, extraversion, agreeableness, and neuroticism. The review begins by analyzing 22 articles from the Science Direct, IEEE Xplore, and Scopus databases by applying screening using the Big Five Personality keywords. Then study the concepts of deep learning for model development such as: input, architecture, method, dataset, and performance to identify the strengths and limitations of each approach. The review results of the highest average accuracy by modality such as unimodal text with essay dataset (80.98%), unimodal image with ImageNet dataset (95.36%), unimodal audio with SSPNet dataset (79.48%), bimodal image and audio with Chalearn Video First Impression dataset (91.67%), multimodal text, image, and audio with Chalearn Video First Impression dataset (91.88%). The average accuracy of the Big Five Personality traits was highest (95.36%) and lowest (59%). This highlights the need for standard benchmarks, interpretable models and interdisciplinary collaboration to improve the accuracy, reliability and ethical implications of personality prediction. Future research is expected to pave the way for advancements in understanding human personality through deep learning approaches using other modalities such as EEG, GSR, and PPG. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Analysis on Performance of Semi-Real-Time Transcription on Health App

Analysis on Performance of Semi-Real-Time Transcription on H...

引用

International Conference on Telecommunication Systems, Services, and Applications (TSSA)

作者： Yoenie Indrasary Daniel Nugraha Ary Setijadi Prihatmanto Ivany Sarief Rudy Gunawan Dhema Yunautama Doctoral Program of Electrical Engineering and Informatics School of Electrical Engineering and Informatics Institut Teknologi Bandung Bandung Indonesia State Polytechnic of Banjarmasin Banjarmasin Indonesia School of Electrical Engineering and Informatics Institut Teknologi Bandung Bandung Indonesia Dept. of Electrical Engineering Faculty of Engineering Universitas Sangga Buana Bandung Indonesia Dept. of Informatics Engineering Directorate of Vocational Universitas Sangga Buana Bandung Indonesia

ISBN: (数字)9798331527686

ISBN: (纸本)9798331527693

The fragmentation of health information systems has hindered seamless data exchange in healthcare services, posing challenges in achieving efficient, real-time access to patient health data. This study addresses these issues by evaluating the implementation of a semi-real-time transcription system using a distributed architecture with a message broker to enhance the documentation of clinical examination processes in primary health services. The research focuses on assessing the feasibility of this approach by measuring key performance indicators such as latency and word error rate (WER) during clinical conversation transcription. Audio recordings were segmented into shorter fragments to simulate real-time transcription, and data transmission was conducted through the Advanced Message Queuing Protocol (AMQP). The findings demonstrate that the average latencys achieved by semi-real-time transcription system is 7.7s, indicating its potential suitability for clinical documentation applications that do not require strict real-time processing. However, glitches observed in the message broker's forwarding mechanism suggest the need for further optimization to ensure consistent performance. Future work aims to develop strategies to minimize transcription errors, enhance reliability, and improve the overall integration of this system in healthcare settings. This study contributes to the advancement of health information systems by offering architecture into scalable and effective methods for semi-real-time data transcription and interoperability.

关键词： Protocols Speech recognition Medical services Documentation Oral communication Real-time systems Telecommunications Reliability Optimization Information systems

来源：评论

学校读者我要写书评

暂无评论

PrimaDNN’:A Characteristics-aware DNN Customization for Singing Technique Detection

arXiv

引用

arXiv 2023年

作者： Yamamoto, Yuya Nam, Juhan Terasawa, Hiroko Doctoral Program in Informatics University of Tsukuba Ibaraki Japan Graduate School of Culture Technology KAIST Daejeon Korea Republic of

Professional vocalists modulate their voice timbre or pitch to make their vocal performance more expressive. Such fluctuations are called singing techniques. Automatic detection of singing techniques from audio tracks can be beneficial to understand how each singer expresses the performance, yet it can also be difficult due to the wide variety of the singing techniques. A deep neural network (DNN) model can handle such variety;however, there might be a possibility that considering the characteristics of the data improves the performance of singing technique detection. In this paper, we propose PrimaDNN, a CRNN model with a characteristics-oriented improvement. The features of the model are: 1) input feature representation based on auxiliary pitch information and multi-resolution mel spectrograms, 2) Convolution module based on the Squeeze-and-excitation (SENet) and the Instance normalization. In the results of J-POP singing technique detection, PrimaDNN achieved the best results of 44.9% at the overall macro-F measure, compared to conventional works. We also found that the contribution of each component varies depending on the type of singing technique. © 2023, CC BY.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Multi-phase Deep Learning Approach for Predicting HCC Response to TACE using Complete Computed Tomography

A Multi-phase Deep Learning Approach for Predicting HCC Resp...

引用

2023 IEEE EMBS Special Topic Conference on Data Science and Engineering in Healthcare, Medicine and Biology, IEEECONF 2023

作者： Chang, Shun Cheng Liu, Chien Fu Misztal, Marta Hsieh, Yi Hsien Lin, Che Taiwan School of Electronic Engineering and Computer Science Queen Mary University of London United Kingdom Department of Electrical Engineering Ntu Taiwan Center for Computational and Systems Biology Center for Biotechnology Ntu Taiwan Smart Medicine and Health Informatics Program Ntu Taiwan

ISBN: (纸本)9798350383386

In this research study, we compare the predictive performance of two advanced deep learning-based models in order to provide a solution to TACE (Transarterial Chemoembolization) response prediction in HCC (Hepatocellular Carcinoma) patients. Using entire abdominal CT scans enabled a broader perspective available for the model, eliminating the need for segmentation during the preprocessing. Making use of both single-phase and multi-phase CT imaging, we have used DenseNet121 and have obtained an accuracy of 80% for the multi-phase *** Relevance: The ability to predict the effectiveness of TACE treatment prior to its administration makes it possible to provide a better decision-making aid for physicians and patients. © 2023 IEEE.

关键词： Computerized tomography

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：