检索结果-内蒙古大学图书馆

31st International Conference on Computational Linguistics, COLING 2025

作者： Sani, Samin Mahdizadeh Sadeghi, Pouya Vu, Thuy-Trang Yaghoobzadeh, Yadollah Haffari, Gholamreza Department of Electrical and Computer Engineering University of Tehran Iran Tehran Institute for Advanced Studies Khatam University Iran Department of Data Science & AI Monash University Australia

ISBN: (纸本)9798891761964

Large language models (LLMs) have made great progress in classification and text generation tasks. However, they are mainly trained on English data and often struggle with low-resource languages. In this study, we explore adding a new language, i.e., Persian, to Llama (a model with a limited understanding of Persian) using parameter-efficient fine-tuning. We employ a multi-stage approach involving pretraining on monolingual Persian data, aligning representations through bilingual pretraining and instruction datasets, and instruction-tuning with task-specific datasets. We evaluate the model's performance at each stage on generation and classification tasks. Our findings suggest that incorporating the Persian language, through bilingual data alignment, can enhance classification accuracy for Persian tasks, with no adverse impact and sometimes even improvements on English tasks. Additionally, the results highlight the model's initial strength as a critical factor when working with limited training data, with cross-lingual alignment offering minimal benefits for the low-resource language. Knowledge transfer from English to Persian has a marginal effect, primarily benefiting simple classification tasks. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Improved Deepmaxoutmodel for Disease Prediction in Healthcare IoT Cloud System 3

Improved Deepmaxoutmodel for Disease Prediction in Healthcar...

引用

3rd IEEE International Conference of Electron Devices Society Kolkata Chapter, EDKCON 2024

作者： Preetha, P. Packialatha, A. Vels Institute of Science Technology and Advanced Studies Department of Computer Science Engineering Tamilnadu Chennai India

ISBN: (纸本)9798350374643

The Internet of Medical Things IoMT has grown rapidly due to the growing popularity of wearable technology and its applications in health monitoring systems. The IoMT significantly reduces the death rate by facilitating early sickness identification. Predicting cardiac disease is among the most significant issues with clinical dataset analysis. Finding the key components of heart disease prediction is the goal of the suggested investigation, which makes use of machine learning techniques. Despite several studies on the issue, the accuracy of the findings on the diagnosis of heart illness is low. The purpose of this paper is to provide an innovative framework for healthcare monitoring that includes the following phases. Data preprocessing, feature extraction, acquisition of data, and prediction are the primary stages. The process of acquiring data is at the IoTs layer. The medical data is considered in this case (benchmark datasets of different illnesses). The data pre-processing operations of data cleaning and data filtering at the cloud layer are applied to the data obtained from the IoT sensors (considering the dataset), where the data normalization process will take place. Subsequently, statistical features and raw features are extracted. The multi-layered Deepmaxout model will predict the diseases during in the prediction phase. Lastly, the evaluation of the suggested approach is compared to existing methods using several types of metrics. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Recognition of Deep Fake Voice Acoustic using Ensemble Bagging Model 5

Recognition of Deep Fake Voice Acoustic using Ensemble Baggi...

引用

5th International Conference on Electronics and Sustainable Communication Systems, ICESC 2024

作者： Basha, Shaik Abdul Yunus Priya, K.Ulaga Vels Institute of Science Technology & Advanced Studies Department of Computer Science and Engineering Tamilnadu Chennai India

ISBN: (纸本)9798350379945

As artificial speech generates technological evolution;fake voice information has become an increasingly prevalent avenue for fraud. Numerous investigations have been carried out into machine learning techniques, signal processing techniques, and the Automatic Speaker Verification System to address such issues. Since there are increasing methods to create fake information, techniques must be developed to determine when audio recordings employed as electronic proof are corrupted. Such recordings contain the potential to be manipulated. Hence, this research work identified the data regarding audio systems as either fake or real by i) extracting features from audio file metadata using oversampling techniques and ii) Developing an Ensemble Bagging Model appropriate for analyzing input audio signal followed by classification of fake (fraudulent) and real (authentic) audio. Our experimental results reveal that the ensemble bagging model attained an accuracy of 99.5%, precision of 99%, 99.6% recall, and 99.9% Fl-measure in the prediction of the audio system compared with traditional approaches. © 2024 IEEE.

关键词： Audio recordings

来源：评论

学校读者我要写书评

暂无评论

An Enhanced Detection System of Autism Spectrum Disorder Using Thermal Imaging and Deep Learning

引用

SN computer Science 2025年第2期6卷 1-11页

作者： Jegan Amarnath, J. Meera, S. Department of Computer Science and Engineering Vels Institute of Science Technology & Advanced Studies (VISTAS) Pallavaram Tamil Nadu Chennai 600117 India

Autism Spectrum Disorder (ASD) is a brain-based condition characterized by social difficulties and repetitive behaviors. Traditional diagnostic methods can be subjective and time-consuming. Current diagnostic techniques for ASD often rely on behavioral assessments and clinical observations, which can be inconsistent and prone to subjectivity. Additionally, there is a need for more reliable and objective methods to differentiate between autistic and non-autistic individuals. Existing imaging techniques require further refinement to improve accuracy and efficiency in detecting ASD. The primary objective of this research is to develop and evaluate an enhanced detection system for ASD using thermal imaging and Improved Mask Faster Recurrent Convolutional Neural Network (IMFRCNN). This research focuses on enhancing the detection of ASD using Deep Learning (DL) methods and thermal imaging. Traditional diagnostic techniques for ASD often rely on subjective behavioral assessments, highlighting the need for more reliable methods. By examining cerebral imaging data from the thermal face image datasets collect facial thermal images using thermal cameras. Ensure the dataset is diverse, including varying expressions and environmental conditions. The study compares the effectiveness of IMFRCNN and the ResNet 50 system in classifying thermal images of 50 autistic and 50 non-autistic individuals. Emotions were evoked using audio-visual stimuli, and temperature variations in facial areas such as the nose, forehead, cheek, and eyes were measured. Anger showed the greatest temperature differences, with a 12.6% variation between autistic and non-autistic individuals. The IMFRCNN achieved a reliability of 96%, while the ResNet 50 achieved 90%. These findings indicate that computer-aided assessment techniques utilizing thermal imaging and DL can provide a consistent and objective method for identifying individuals with ASD. © The Author(s), under exclusive licence to Springer Nature Sing

关键词： Deep learning algorithms Diagnostic tool Improved mask faster recurrent convolutional neural network Performance measures

来源：评论

学校读者我要写书评

暂无评论

Optimized Deep Learning Approach for Efficient Diabetic Retinopathy Classification Combining VGG16-CNN

引用

computers, Materials & Continua 2023年第11期77卷 1855-1872页

作者： Heba M.El-Hoseny Heba F.Elsepae Wael A.Mohamed Ayman S.Selmy Department of Computer Science the Higher Future Institute for Specialized Technological StudiesObour11828Egypt Department of Electrical Engineering Benha Faculty of EngineeringBenha UniversityBenha13511Egypt

Diabetic retinopathy is a critical eye condition that,if not treated,can lead to vision *** methods of diagnosing and treating the disease are time-consuming and ***,machine learning and deep transfer learning(DTL)techniques have shown promise in medical applications,including detecting,classifying,and segmenting diabetic *** advanced techniques offer higher accuracy and *** Diagnosis(CAD)is crucial in speeding up classification and providing accurate disease ***,these technological advancements hold great potential for improving the management of diabetic *** study’s objective was to differentiate between different classes of diabetes and verify the model’s capability to distinguish between these *** robustness of the model was evaluated using other metrics such as accuracy(ACC),precision(PRE),recall(REC),and area under the curve(AUC).In this particular study,the researchers utilized data cleansing techniques,transfer learning(TL),and convolutional neural network(CNN)methods to effectively identify and categorize the various diseases associated with diabetic retinopathy(DR).They employed the VGG-16CNN model,incorporating intelligent parameters that enhanced its *** outcomes surpassed the results obtained by the auto enhancement(AE)filter,which had an ACC of over 98%.The manuscript provides visual aids such as graphs,tables,and techniques and frameworks to enhance *** study highlights the significance of optimized deep TL in improving the metrics of the classification of the four separate classes of *** manuscript emphasizes the importance of using the VGG16CNN classification technique in this context.

关键词： No diabetic retinopathy(NDR) convolution layers(CNV layers) transfer learning data cleansing convolutional neural networks a visual geometry group(VGG16)

来源：评论

学校读者我要写书评

暂无评论

Benchmarking Large Language Models for Persian: A Preliminary Study Focusing on ChatGPT 30

Benchmarking Large Language Models for Persian: A Preliminar...

引用

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

作者： Abaskohi, Amirhossein Baruni, Sara Masoudi, Mostafa Abbasi, Nesa Babalou, Mohammad Hadi Edalat, Ali Kamahi, Sepehr Sani, Samin Mahdizadeh Naghavian, Nikoo Namazifard, Danial Sadeghi, Pouya Yaghoobzadeh, Yadollah Computer Science Department University of British Columbia Vancouver Canada School of Electrical and Computer Engineering College of Engineering University of Tehran Tehran Iran Tehran Institute for Advanced Studies Khatam University Tehran Iran

ISBN: (纸本)9782493814104

This paper explores the efficacy of large language models (LLMs) for Persian. While ChatGPT and consequent LLMs have shown remarkable performance in English, their efficiency for more low-resource languages remains an open question. We present the first comprehensive benchmarking study of LLMs across diverse Persian language tasks. Our primary focus is on GPT-3.5-turbo, but we also include GPT-4 and OpenChat-3.5 to provide a more holistic evaluation. Our assessment encompasses a diverse set of tasks categorized into classic, reasoning, and knowledge-based domains. To enable a thorough comparison, we evaluate LLMs against existing task-specific fine-tuned models. Given the limited availability of Persian datasets for reasoning tasks, we introduce two new benchmarks: one based on elementary school math questions and another derived from the entrance exams for 7th and 10th grades. Our findings reveal that while LLMs, especially GPT-4, excel in tasks requiring reasoning abilities and a broad understanding of general knowledge, they often lag behind smaller pretrained models fine-tuned specifically for particular tasks. Additionally, we observe improved performance when test sets are translated to English before inputting them into GPT-3.5. These results highlight the significant potential for enhancing LLM performance in the Persian language. This is particularly noteworthy due to the unique attributes of Persian, including its distinct alphabet and writing styles. We have made our codes, prompts, and data available here: https://***/Ipouyall/Benchmarking_ChatGPT_for_Persian. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Counterfactuals As a Means for Evaluating Faithfulness of Attribution Methods in Autoregressive Language Models 7

Counterfactuals As a Means for Evaluating Faithfulness of At...

引用

7th BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, BlackboxNLP 2024

作者： Kamahi, Sepehr Yaghoobzadeh, Yadollah School of Electrical and Computer Engineering College of Engineering University of Tehran Tehran Iran Tehran Institute for Advanced Studies Khatam University Tehran Iran

ISBN: (纸本)9798891761704

Despite the widespread adoption of autoregressive language models, explainability evaluation research has predominantly focused on span infilling and masked language models. Evaluating the faithfulness of an explanation method—how accurately it explains the inner workings and decision-making of the model—is challenging because it is difficult to separate the model from its explanation. Most faithfulness evaluation techniques corrupt or remove input tokens deemed important by a particular attribution (feature importance) method and observe the resulting change in the model’s output. However, for autoregressive language models, this approach creates out-of-distribution inputs due to their next-token prediction training objective. In this study, we propose a technique that leverages counterfactual generation to evaluate the faithfulness of attribution methods for autoregressive language models. Our technique generates fluent, in-distribution counterfactuals, making the evaluation protocol more reliable. ©2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Machine-learned wearable sensors for real-time hand-motion recognition: toward practical applications

引用

National Science Review 2024年第2期11卷 231-263页

作者： Kyung Rok Pyun Kangkyu Kwon Myung Jin Yoo Kyun Kyu Kim Dohyeon Gong Woon-Hong Yeo Seungyong Han Seung Hwan Ko Department of Mechanical Engineering Seoul National University IEN Center for Human-Centric Interfaces and Engineering Institute for Electronics and NanotechnologyGeorgia Institute of Technology School of Electrical and Computer Engineering Georgia Institute of Technology Department of Chemical Engineering Stanford University Department of Mechanical Engineering Ajou University George W.Woodruff School of Mechanical Engineering Georgia Institute of Technology Institute of Advanced Machinery and Design(SNU-IAMD) Seoul National University

Soft electromechanical sensors have led to a new paradigm of electronic devices for novel motion-based wearable applications in our daily lives. However, the vast amount of random and unidentified signals generated by complex body motions has hindered the precise recognition and practical application of this technology. Recent advancements in artificial-intelligence technology have enabled significant strides in extracting features from massive and intricate data sets, thereby presenting a breakthrough in utilizing wearable sensors for practical applications. Beyond traditional machine-learning techniques for classifying simple gestures, advanced machine-learning algorithms have been developed to handle more complex and nuanced motion-based tasks with restricted training data sets. Machine-learning techniques have improved the ability to perceive, and thus machine-learned wearable soft sensors have enabled accurate and rapid human-gesture recognition, providing real-time feedback to users. This forms a crucial component of future wearable electronics, contributing to a robust human–machine interface. In this review, we provide a comprehensive summary covering materials, structures and machine-learning algorithms for hand-gesture recognition and possible practical applications through machine-learned wearable electromechanical sensors.

关键词： wearable sensor soft electronics artificial intelligence machine learning human–machine interfaces gesture recognition

来源：评论

学校读者我要写书评

暂无评论

Active pixel image sensor array for dual vision using large-area bilayer WS_(2)

引用

InfoMat 2024年第4期6卷 68-82页

作者： Arindam Bala Mayuri Sritharan Na Liu Muhammad Naqi Anamika Sen Gyuchull Han Hyun Yeol Rho Youngki Yoon Sunkook Kim School of Advanced Materials Science and Engineering Sungkyunkwan UniversitySuwonRepublic of Korea Waterloo Institute for Nanotechnology(WIN) University of WaterlooWaterlooOntarioCanada Department of Electrical and Computer Engineering University of WaterlooWaterlooOntarioCanada

Transition metal dichalcogenides(TMDs)are a promising candidate for developing advanced sensors,particularly for day and night vision systems in vehicles,drones,and security *** traditional systems rely on separate sensors for different lighting conditions,TMDs can absorb light across a broad-spectrum *** this study,a dual vision active pixel image sensor array based on bilayer WS2 phototransistors was *** bilayer WS2 film was synthesized using a combined process of radio-frequency sputtering and chemical vapor *** WS2-based thin-film transistors(TFTs)exhibit high average mobility,excellent Ion/Ioff,and uniform electrical *** optoelectronic properties of the TFTs array exhibited consistent behavior and can detect visible to near-infrared light with the highest responsivity of 1821 A W1(at a wavelength of 405 nm)owing to the photogating ***,red,green,blue,and near-infrared image sensing capabilities of active pixel image sensor array utilizing light stencil projection were *** proposed image sensor array utilizing WS2 phototransistors has the potential to revolutionize the field of vision sensing,which could lead to a range of new opportunities in various applications,including night vision,pedestrian detection,various surveillance,and security systems.

关键词： dual vision image sensors infrared wavelength detection phototransistors thin-film transistor arrays transition metal dichalcogenides

来源：评论

学校读者我要写书评

暂无评论

A Robotic Arm Vehicle using Voice Recognition for Physically Challenged People 2

A Robotic Arm Vehicle using Voice Recognition for Physically...

引用

2nd International Conference on Advances in Computation, Communication and Information Technology, ICAICCIT 2024

作者： Nannda, Jia Dora, Lokareddy Venkanna Gupta, Tanvi Chouhan, Aryan Verma, Anmol Department of Computer Science and Engineering Manav Rachna International Institute of Research and Studies India Department of Computer Science and Engineering Manav Rachna International Institute of Research and Studies Faridabad India Department of electrical and electronics engineering Manav Rachna International Institute of Research and Studies India

ISBN: (纸本)9798331541217

In Today's world, all the work is automated by robots. The Robots have become a significant part of today's world, improving many people's lives, especially those who are physically challenged. This research paper presents a unique method for helping people with physical disabilities: A robotic arm vehicle using voice recognition for physically challenged people. Two remarkable features of the robot are its design and its adaptability in handling objects accurately. The comparative research demonstrates improved performance, such as faster job completion and increased grasping efficiency, when compared to older methods. Additionally, by integrating AI, the robotic arm gains the flexibility needed to operate in real-world environments. This finding might lead to more autonomous and effective assistive technology for personal support and healthcare applications. © 2024 IEEE.

关键词： Robotic arms

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：