检索结果-内蒙古大学图书馆

2nd International Conference on Data science and Network Engineering, ICDSNE 2024

作者： Bandopadhyay, Saikat Dutta, Jhalak Das, Smita Datta, Soumyajit Ghosh, Debaditya Dey, Rohit Kumar Nandigrami, Jeet Department of Artificial Intelligence and Machine Learning Netaji Subhash Engineering College Kolkata India Department of Computer Science and Engineering Heritage Institute of Technology Kolkata India Department of Computer Science and Engineering National Institute of Technology Agartala India Department of Artificial Intelligence and Machine Learning Heritage Institute of Technology Kolkata India

ISBN: (纸本)9789819783359

Blindness or visual impairment ranks among the top ten disabilities affecting both men and women, impacting over 35 million individuals of all ages in India. Access to visual information is crucial for enhancing the independence and safety of blind and visually impaired individuals. There exists a compelling imperative to develop assistive technologies that can significantly improve their quality of life. This paper presents SightAssist, a multilingual software solution designed to address the challenges faced by the visually impaired community in India. SightAssist utilizes a camera-based approach to serve as a digital prosthetic eye and virtual assistant, capable of currency detection, facial recognition, and providing various functionalities to assist with daily activities. Initially, a custom-labeled dataset has been prepared using a pre-trained YOLO model. Gradually, dataset pre-processing, normalization, and semantic segmentation are employed to optimize model performance. Additionally, voice-based accessibility features have been integrated using the Web Speech API, enabling users to interact with the application through voice commands. The experimental results show the successful development of SightAssist and its potential to significantly improve the lives of visually impaired people. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Assistive technology

来源：评论

学校读者我要写书评

暂无评论

OncoNLP: Cancer Comprehend Annotation - A Pipeline for Cancer Phenotype and Clinical Extraction

OncoNLP: Cancer Comprehend Annotation - A Pipeline for Cance...

引用

2024 IEEE-EMBS International Conference on Biomedical and Health Informatics, BHI 2024

作者： Duong, Thanh Thieu, Thanh University of South Florida Department of Computer Science and Engineering TampaFL United States Moffitt Cancer Center Department of Machine Learning TampaFL United States

ISBN: (纸本)9798350351552

Information extraction from clinical text is needed to comprehend patient conditions and determine anticancer treatment. Existing NLP systems, such as Amazon Comprehend Medical, Clamp, and DeepPhe, are proprietary and suboptimal due to the shift in the textual distribution and expression of cancer phenotypes. We introduce OncoNLP, a natural language processing toolkit that comprises deep neural network BERT models designed to extract cancer phenotype and related biomedical information. Currently, the toolkit contains two primary components: a Biomedical BERT(BiomedBERT) model to extract general medical information and a Cancer Bert(CaBERT) model to identify the primary tumor site and histology. We evaluate the performance of BiomedBERT on Informatics for Integrating Biology & the Bedside (i2b2) dataset against Amazon Comprehend Medical and Clamp. BiomedBERT outperforms all other methods with an exact matching F1-score of 88.5%. Next, we evaluate CaBERT on a Moffitt dataset with over 2000 clinical notes against DeepPhe, which shows a 50% improvement in both tumor site and histology tasks. Lastly, we introduce a knowledge prompt with OncoNLP and engineer it for large language models. We name it OncoNLP-Assist, a chatbot system powered by OncoNLP and Llama2 that could extract information from electrical health records and interact with physicians. © 2024 IEEE.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

引用

2023 International Conference on Artificial Intelligence science and Applications in Industry and Society, CAISAIS 2023

作者： Ghanem, Sahar I. Computer Science and Artificial Intelligence Faculty Artificial Intelligence and Machine Learning Department Egypt

ISBN: (纸本)9798350314786

In an attempt to facilitate the way students, succeed academically, academic advising is an integral part of the educational system. However, it has frequently been dependent on time-consuming manual processes and personalized interaction, which are not scalable. machine learning and deep learning have the potential to completely transform advising by automating parts of process and giving students distinctive guidance. This paper explores the relationship between student academic performance and correlated features using deep learning methodology in academic advising systems. The model was trained and tested with the Kaggle dataset, and the results show high accuracy, specificity, and true positive rate values compared to naive bayes and random forest methodologies. The paper also addresses potential challenges and limitations of deep learning in academic advising, such as data privacy concerns, algorithmic bias, and the need for human intervention. It concludes by emphasizing the importance of incorporating ethical considerations and human expertise in the development and deployment of deep learning models in academic advising. Overall, the paper highlights the potential of deep learning algorithms to transform academic advising and improve student success. © 2023 IEEE.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

SecureFedPROM: A Zero-Trust Federated learning Approach with Multi-Criteria Client Selection

引用

IEEE Journal on Selected Areas in Communications 2025年第6期43卷 2025-2041页

作者： Tahir, Mehreen Mawla, Tanjila Awaysheh, Feras Alawadi, Sadi Gupta, Maanak Ali, Muhammad Intizar Dublin City University SFI Centre for Research Training in Machine Learning Dublin Ireland Tennessee Tech University Department of Computer Science CookevilleTN United States Umeå University Department of Computer Science Umeå Sweden Blekinge Tekniska Högskola School of Computer Science Karlskrona Sweden Dublin City University School of Electronic and Computer Engineering Dublin Ireland

Federated learning (FL) enables decentralized learning while preserving data privacy. However, ensuring security and optimizing resource utilization in FL remains challenging, particularly in untrusted environments. To address this, we propose SecureFedPROM, a novel zero-trust FL framework that integrates Attribute-Based Access Control (ABAC) for secure client authorization and Preference Ranking Organization Method for Enrichment of Evaluations (PROMETHEE) for dynamic, multi-criteria client selection. Unlike traditional FL client selection methods that prioritize security or efficiency, SecureFedPROM optimizes trustworthiness, computational efficiency, and performance, ensuring robust participation in each training round. We evaluate SecureFedPROM across multiple real-world datasets, demonstrating its superiority over state-of-the-art client selection protocols. Our results show that SecureFedPROM achieves a 7.19% improvement in model accuracy, accelerates convergence, and reduces the number of training rounds. Additionally, it minimizes wall-clock time and computational overhead, making it highly scalable for edge AI environments. These findings highlight the importance of integrating zero-trust security principles with multi-criteria decision-making to enhance security and efficiency in FL. © 1983-2012 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Image Super Resolution using U-Net architecture and SRGAN: Comparative Analysis 2

Image Super Resolution using U-Net architecture and SRGAN: C...

引用

2nd International Conference on computer, Communication and Control, IC4 2024

作者： Dandekar, Pranali Bhojwani, Bhavika Balpande, Aarya Zanwar, Sanskar Deb, Ankan Shri Ramdeobaba College of Engineering and Management Department of Computer Science and Engineering [Artificial Intelligence and Machine Learning Nagpur India

ISBN: (纸本)9798350387933

Super-resolution is an important image processing task that improves the resolution of high-resolution images. Low resolution is used in many different fields, including medical imaging, satellite imagery, and computer vision. This paper presents a new approach to ultra-high resolution using the U-Net architecture, a deep learning framework known for its success in image segmentation and restoration *** this study, we propose an adaptive U-Net model specifically designed for ultra-high-resolution tasks. The architecture includes an encoder-decoder network with bypass connections, enabling multi-scale feature extraction and high-resolution detail reconstruction. Our model is trained on a diverse dataset of low- and high-resolution image pairs, allowing it to learn complex relationships and patterns in *** evaluate the performance of the U-Net based superresolution method using standard image quality metrics and qualitative visual evaluation. Test results show a significant improvement in image quality, with improved sharpness, texture, and detail recovery. Furthermore, our model outperforms state-of-the-art super-resolution methods in terms of signal-to-noise ratio (PSNR) and structural similarity index (SSIM). © 2024 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Enhancement of Infant Health Assessment: Predicting Body Mass Index (BMI) from Real-Time Facial Images Using machine learning Techniques 23th

Enhancement of Infant Health Assessment: Predicting Body Mas...

引用

23rd International Conference on Intelligent Systems Design and Applications, ISDA 2023

作者： Krishnaveni, M. Subashini, P. Janani, R. Jeeva, N. Department of Computer Science Centre for Machine Learning and Intelligence Avinashilingam Institute for Home Science and Higher Education for Women Coimbatore India Centre for Machine Learning and Intelligence Avinashilingam Institute for Home Science and Higher Education for Women Coimbatore India Department of Computer Science Avinashilingam Institute for Home Science and Higher Education for Women Coimbatore India

ISBN: (纸本)9783031648120

Body Mass Index in infants is a valuable indicator for assessing their growth and nutritional state, helping to identify any potential issues in the early stages. It establishes whether infants are underweight, within a healthy weight range, or overweight, which can specify their overall health. The capability to accurately determine an infant’s BMI can provide significant insight into their general health and helps parents to adopt smart feeding and healthcare decisions. In today’s world, where parents often have limited time and wealth, supporting the proper growth of their infant is vital. The fundamental objective of this research is to propose a user-friendly mobile application which is constructed for predicting infant BMI. This proposed application provides a simple and easy way for parents to track and understand their infant’s nutritionary condition. The proposed application requires only a facial image of an infant, eliminating the necessity for addition parameters such as height, weight. For the prediction of infant BMI, regression algorithms, which includes Linear, Polynomial, Random Forest, and Kernal Ridge, have been employed. The performance of these models has been assessed using several evaluation metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared (R2) score, and the results clearly show their effectiveness in this significant task. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： machine learning

来源：评论

学校读者我要写书评

暂无评论

Typos Correction Training Against Misspellings from Text-to-Text Transformers 30

Typos Correction Training Against Misspellings from Text-to-...

引用

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Xie, Guicai Zhang, Ke Duan, Lei Zhang, Wei Huang, Zeqian Machine Learning Platform Department Tencent Inc. China School of Computer Science Sichuan University Chengdu China

ISBN: (纸本)9782493814104

Dense retrieval (DR) has become a mainstream approach to information seeking, where a system is required to return relevant information to a user query. In real-life applications, typoed queries resulting from the users' mistyping words or phonetic typing errors exist widely in search behaviors. Current dense retrievers experience a significant drop in retrieval effectiveness when they encounter typoed queries. Therefore, the search system requires the extra introduction of spell-checkers to deal with typos and then applies the DR model to perform robust matching. Herein, we argue that directly conducting the typos correction training would be beneficial to make an end-to-end retriever against misspellings. To this end, we propose a novel approach that can facilitate the incorporation of the spelling correction objective into the DR model using the encoder-decoder architecture. During typos correction training, we also develop a prompt-based augmentation technique to enhance the DR space alignment of the typoed query and its original query. Extensive experiments demonstrate that the effectiveness of our proposed end-to-end retriever significantly outperforms existing typos-aware training approaches and sophisticated training advanced retrievers. Our code is available at https://***/striver314/ToCoTR. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Search engines

来源：评论

学校读者我要写书评

暂无评论

MS3D: A RG Flow-Based Regularization for GAN Training with Limited Data 41

MS3D: A RG Flow-Based Regularization for GAN Training with L...

引用

41st International Conference on machine learning, ICML 2024

作者： Wang, Jian Lan, Xin Tian, Yuxin Lv, Jiancheng College of Computer Science Sichuan University Engineering Research Center of Machine Learning and Industry Intelligence Ministry of Education Chengdu610065 China

Generative adversarial networks (GANs) have made impressive advances in image generation, but they often require large-scale training data to avoid degradation caused by discriminator overfitting. To tackle this issue, we investigate the challenge of training GANs with limited data, and propose a novel regularization method based on the idea of renormalization group (RG) in *** observe that in the limited data setting, the gradient pattern that the generator obtains from the discriminator becomes more aggregated over time. In RG context, this aggregated pattern exhibits a high discrepancy from its coarse-grained versions, which implies a high-capacity and sensitive system, prone to overfitting and collapse. To address this problem, we introduce a multi-scale structural self-dissimilarity (MS3D) regularization, which constrains the gradient field to have a consistent pattern across different scales, thereby fostering a more redundant and robust system. We show that our method can effectively enhance the performance and stability of GANs under limited data scenarios, and even allow them to generate high-quality images with very few data. Copyright 2024 by the author(s)

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Adversarial learning Based Knowledge Distillation on 3D Point Clouds

Adversarial Learning Based Knowledge Distillation on 3D Poin...

引用

2025 IEEE/CVF Winter Conference on Applications of computer Vision, WACV 2025

作者： Sanjay, S.J. Akash, J. Rajan, Sreehari Shajahan, Dimple A Sharma, Charu Iiit Hyderabad Machine Learning Lab India Tkm College of Engineering Department of Computer Science Kollam India

ISBN: (纸本)9798331510831

The significant improvements in point cloud representation learning have increased its applicability in many real-life applications, resulting in the need for lightweight, better-performing models. One widely proposed efficient method is knowledge distillation, where a lightweight model uses knowledge from large models. Very few works exist on distilling the knowledge for point clouds. Most of the work focuses on cross-modal-based approaches that make the method expensive to train. This paper proposes PointKAD, an adversarial knowledge distillation framework for point cloud-based tasks. PointKAD includes adversarial feature distillation and response distillation with the help of discriminators to extract and distill the representation of feature maps and logits. We conduct extensive experimental studies on both synthetic (ModelNet40) and real (ScanObjectNN) datasets to show that PointKAD achieves state-of-the-art results compared to the existing knowledge distillation methods for point cloud classification. Additionally, we present results on the part segmentation task, highlighting the efficacy of the PointKAD framework. Our experiments further reveal that PointKAD is capable of transferring knowledge across different tasks and datasets, show-casing its versatility. Furthermore, we demonstrate that PointKAD can be applied to a cross-modal training setup, achieving competitive performance with cross-modal-based point cloud methods for classification. © 2025 IEEE.

关键词： Modal analysis

来源：评论

学校读者我要写书评

暂无评论

Understanding Finetuning for Factual Knowledge Extraction 41

Understanding Finetuning for Factual Knowledge Extraction

引用

41st International Conference on machine learning, ICML 2024

作者： Ghosal, Gaurav Hashimoto, Tatsunori Raghunathan, Aditi Department of Machine Learning Carnegie Mellon University Pittsburgh United States Department of Computer Science Stanford University Stanford United States

In this work, we study the impact of QA finetuning data on downstream factuality. We show that fine-tuning on lesser-known facts that are poorly stored during pretraining yields significantly worse factuality than fine-tuning on well-known facts, even when all facts are seen during pretraining. We prove this phenomenon theoretically, showing that training on lesser-known facts can lead the model to ignore subject entity names and instead output a generic plausible response even when the relevant factual knowledge is encoded in the model. On three question answering benchmarks (PopQA, Entity Questions, and MMLU) and two language models (Llama-2-7B and Mistral-7B), we find that (i) finetuning on a completely factual but lesser-known subset of the data deteriorates downstream factuality (5-10%) and (ii) finetuning on a subset of better-known examples matches or outperforms finetuning on the entire dataset. Ultimately, our results shed light on the interaction between pretrained knowledge and finetuning data and demonstrate the importance of taking into account how facts are stored in the pretrained model when fine-tuning for knowledge-intensive tasks. Copyright 2024 by the author(s)

关键词： Question answering

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：