检索结果-内蒙古大学图书馆

Deep Learning for Detecting Mental Disorder from User Content on Social Media

SN Computer Science 2025年第2期6卷 1-29页

作者： Tounsi, Malek Ameur, Hanen Jamoussi, Salma Multimedia Information Systems and Advanced Computing Laboratory University of Sfax Technopole of Sfax Sfax Tunisia

The prevalence of mental health disorders is rising, with many individuals expressing their thoughts and feelings through social media posts. This presents an opportunity for early detection of mental conditions using natural language processing (NLP). This paper addresses the lack of effective tools for detecting mental disorders from Arabic social media content. We propose Ara-MentalBERT, a novel deep learning fusion model that combines AraBERT and MentalBERT using a deep joint autoencoder (DJAE). The DJAE model aims to integrate Arabic language understanding with specialized mental health domain knowledge by fusing Arabic and mental health embedding representations of text. Our proposed method involves three key steps: (1) fine-tuning pre-trained AraBERT and MentalBERT models, (2) employing a deep joint autoencoder for Ara-Mental text representation learning, and (3) utilizing a mental health disorder classifier with the fused Ara-Mental representations. We evaluate our model on two diverse datasets: a Reddit dataset and a Twitter dataset. Results demonstrate significant improvements over existing methods, achieving an accuracy of 81.12% on the Reddit dataset for combined titles and posts and 86.12% accuracy on the Arabic tweets. These findings highlight the potential of our model to enhance mental health detection in Arabic-speaking communities, offering a powerful tool for identifying at-risk individuals more effectively. Our proposed method makes valuable progress at the intersection of NLP, mental health, and deep learning, paving the way for Arabic social media monitoring to facilitate early intervention and improve mental healthcare. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Ara-Mental representation Ara-MentalBERT AraBERT Deep joint autoencoder Deep learning Mental health disorders MentalBERT

来源：评论

学校读者我要写书评

暂无评论

Hybrid EEG-fNIRS Detection of MCI Subtypes Based on Transformer Network

引用

SN Computer Science 2025年第4期6卷 1-13页

作者： Bouaziz, Bassem Chaabene, Siwar Mahdi, Walid Multimedia InfoRmation Systems and Advanced Computing Laboratory (MIRACL) University of Sfax Sfax 3023 Tunisia

Preventing the progression of mild cognitive impairment (MCI) to dementia is of paramount importance in the field of healthcare. Furthermore, detecting the subtle indications of MCI in its early stages is difficult but necessary. MCI can appear as either amnestic (aMCI) or non-amnestic (naMCI), confounding the diagnosis even more. However, with the development of improved neuroimaging techniques, computational algorithms that analyze data acquired during clinical tests might provide a solution. Among them, functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) offer invaluable insights into an individual’s cognitive state. In this paper, we present a novel approach for distinguishing individuals diagnosed with MCI, specifically aMCI and naMCI, from healthy controls (HC). We employ a hybrid fusion of EEG and fNIRS signals, acquired during a cognitive task. To address this complex classification problem, we propose a new method based on the transformer network for MCI detection. Our experimental results demonstrate a remarkable classification accuracy of 99.78%, highlighting the significant improvement in MCI detection capabilities achieved through the integration of multimodal neuroimaging data. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Cognitive task EEG FNIRS HC vs aMCI vs naMCI MCI detection Multi-modality Transformer network

来源：评论

学校读者我要写书评

暂无评论

Parameter-efficient fine-tuning of Whisper for low-resource speech recognition 5

Parameter-efficient fine-tuning of Whisper for low-resource ...

引用

5th International Seminar on Artificial Intelligence, Networking and information Technology, AINIT 2024

作者： Liu, Yunpeng Qu, Dan School of Information Systems Engineering University of Information Engineering Laboratory For Advanced Computing and Intelligence Engineering Department of Artificial Intelligence Henan China Systems Engineering University of Information Engineering Laboratory For Advanced Computing and Intelligence Engineering Henan China

ISBN: (纸本)9798350385557

Limited data availability remains a significant challenge for Whisper's low-resource speech recognition performance, falling short of practical application requirements. While previous studies have successfully reduced the recognition error rates of target language speech through fine-tuning, a comprehensive exploration and analysis of Whisper's parameter-efficient fine-tuning capabilities, the advantages and disadvantages are still lacking. This paper aims to fill this gap by conducting comprehensive experimental exploration for Whisper's low-resource speech recognition performance using two parameter-efficient fine-tuning strategies with limited supervised data from multiple low-resource languages. The results and analysis demonstrate that all fine-tuning strategies explored in this paper significantly enhance Whisper's performance. However, different strategies vary in their suitability and practical effectiveness, highlighting the need for careful selection based on specific use cases and resources available. © 2024 IEEE.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Meta-Adapter for Self-Supervised Speech Models: A Solution to Low-Resource Speech Recognition Challenges 30

Meta-Adapter for Self-Supervised Speech Models: A Solution t...

引用

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

作者： Chen, Yaqi Zhang, Hao Yang, Xukui Zhang, Wenlin Qu, Dan School of Information Systems Engineering Information Engineering University China Laboratory for Advanced Computing and Intelligence Engineering China

ISBN: (纸本)9782493814104

Self-supervised models have demonstrated remarkable performance in speech processing by learning latent representations from large amounts of unlabeled data. Although these models yield promising results on low-resource languages, the computational expense of fine-tuning all model parameters is prohibitively high. Adapters offer a solution by incorporating lightweight bottleneck structures into pre-trained models, enabling efficient parameter adaptation for downstream tasks. However, randomly initialized adapters often underperform in low-resource scenarios, limiting their applicability in low-resource languages. To address this issue, we develop the Meta-Adapter for self-supervised models to obtain meta-initialized parameters that facilitate quick adaptation to low-resource languages. Extensive experiments on the Common Voice and FLEURS datasets demonstrate the superior performance of Meta-Adapters on 12 low-resource languages spanning four different language families. Moreover, Meta-adapters show better generalization and extensibility than traditional pretraining methods. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Speech processing

来源：评论

学校读者我要写书评

暂无评论

Detection of Heart Diseases Using CNN-LSTM 22nd

Detection of Heart Diseases Using CNN-LSTM

引用

22nd International Conference on Hybrid Intelligent systems, HIS 2022 and the 18th International Conference on information Assurance and Security, IAS 2022

作者： Karoui, Hend Hamza, Sihem Ayed, Yassine Ben Multimedia Information Systems and Advanced Computing Laboratory MIRACL University of Sfax Sfax Tunisia

ISBN: (纸本)9783031274084

The ElectroCardioGram (ECG) is one of the most used signals for the prediction of heart disease. Lots of research are based on the ECG signal for the detection of cardiac diseases. In this research, we propose a four phases method to the recognition of cardiac disease. The first phase is to remove noise and detect QRS complex using a pass-band filter. In the second phase, we segment the filtered signal. For third phase is to make a fusion of three types of characteristics such as Zero Crossing Rate (ZCR), entropy and cepstral coefficients. Those features extracted considered as input in the next step which is the combination between Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) proposed for multi-class classification of the ECG signal into five classes: Normal beat (N), Left bundle-branch block beat (L), Right bundle-branch block beat (R), premature Ventricular contraction (V), and Paced beat (P). The proposed model was evaluated using the MITBIH arrhythmia database and achieve an accuracy equal to 95.80%. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

Opportunities for retrieval and tool augmented large language models in scientific facilities

引用

npj Computational Materials 2024年第1期10卷 588-595页

作者： Michael H.Prince Henry Chan Aikaterini Vriza Tao Zhou Varuni K.Sastry Yanqi Luo Matthew T.Dearing Ross J.Harder Rama K.Vasudevan Mathew J.Cherukara Advanced Photon Source Argonne National LaboratoryLemontILUSA Center for Nanoscale Materials Argonne National LaboratoryLemontILUSA Argonne Leadership Computing Facility Argonne National LaboratoryLemontILUSA Business and Information Systems Argonne National LaboratoryLemontILUSA Center for Nanophase Materials Oak Ridge National LaboratoryOak RidgeTNUSA

Upgrades to advanced scientific user facilities such as next-generation x-ray light sources,nanoscience centers,and neutron facilities are revolutionizing our understanding of materials across the spectrum of the physical sciences,from life sciences to ***,these facility and instrument upgrades come with a significant increase in *** by more exacting scientific needs,instruments and experiments become more intricate each *** increased operational complexity makes it ever more challenging for domain scientists to design experiments that effectively leverage the capabilities of and operate on these advanced *** language models(LLMs)can perform complex information retrieval,assist in knowledge-intensive tasks across applications,and provide guidance on tool *** x-ray light sources,leadership computing,and nanoscience centers as representative examples,we describe preliminary experiments with a Context-Aware Language Model for Science(CALMS)to assist scientists with instrument operations and complex *** the ability to retrieve relevant information from facility documentation,CALMS can answer simple questions on scientific capabilities and other operational *** the ability to interface with software tools and experimental hardware,CALMS can conversationally operate scientific *** making information more accessible and acting on user needs,LLMs could expand and diversify scientific facilities’users and accelerate scientific output.

关键词： instrument instruments facilities

来源：评论

学校读者我要写书评

暂无评论

Improving Speech Emotion Recognition Using Data Augmentation and Balancing Techniques 22

Improving Speech Emotion Recognition Using Data Augmentation...

引用

22nd International Conference on Cyberworlds, CW 2023

作者： Barhoumi, Chawki Ayed, Yassine Ben National School of Electronics and Telecommunications of Sfax Stax Tunisia Multimedia InfoRmation Systems and Advanced Computing Laboratory Sfax Tunisia

ISBN: (纸本)9798350315653

Speech Emotion Recognition (SER) is a challenging task due to the complexity and variability of human emotions. In this paper, we propose an innovative approach to improve SER performance on the EMODB dataset. Our approach employs data augmentation techniques, such as noise addition and spectrogram shift, as well as balancing techniques, including random oversampling. We also extract five different features from the dataset samples: MFCC, Chroma, Mel Spectrogram, ZCR, and RMS. We compare the performance of four different classifiers MLP, SVM, KNN, and CNN with and without the use of our proposed approach. Our results demonstrate that the proposed approach significantly enhances the accuracy of speech emotion recognition compared to the approach without data augmentation and balancing techniques. Our experiments reveal that the proposed approach achieves higher accuracy and F1-score compared to other approaches, with MLP and CNN models achieving 100% accuracy. These findings highlight the effectiveness of data augmentation and balancing techniques in improving the performance of speech emotion recognition. Moreover, our approach holds great potential for application in various real-life scenarios, including mental health monitoring, human-robot interaction, and speech-based virtual assistants. © 2023 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Adaptive Time Slot Resource Allocation in SWIPT IoT Networks

引用

Computer Modeling in Engineering & Sciences 2023年第9期136卷 2787-2813页

作者： Yunong Yang Yuexia Zhang Zhihai Zhuo Key Laboratory of Information and Communication Systems Ministry of Information IndustryBeijing Information Science and Technology UniversityBeijing100101China Key Laboratory of ModernMeasurement&Control Technology Ministry of EducationBeijing Information Science&Technology UniversityBeijing100101China Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing Beijing100101China

The rapid advancement of Internet of Things(IoT)technology has brought convenience to people’s lives;however further development of IoT faces serious challenges,such as limited energy and shortage of network spectrum *** address the above challenges,this study proposes a simultaneous wireless information and power transfer IoT adaptive time slot resource allocation(SIATS)***,an adaptive time slot consisting of periods for sensing,information transmission,and energy harvesting is designed to ensure that the minimum energy harvesting requirement ismet while the maximumuplink and downlink throughputs are ***,the optimal transmit power and channel assignment of the system are obtained using the Lagrangian dual and gradient descent methods,and the optimal time slot assignment is determined for each IoT device such that the sum of the throughput of all devices is *** results show that the SIATS algorithm performs satisfactorily and provides an increase in the throughput by up to 14.4%compared with that of the fixed time slot allocation(FTS)*** the case of a large noise variance,the SIATS algorithm has good noise immunity,and the total throughput of the IoT devices obtained using the SIATS algorithm can be improved by up to 34.7%compared with that obtained using the FTS algorithm.

关键词： Internet of Things simultaneous wireless information and power transfer adaptive time slot resource allocation

来源：评论

学校读者我要写书评

暂无评论

Bridging the Gap using Contrastive Learning and Semantic Consistency for Unsupervised Neural Machine Translation 5

Bridging the Gap using Contrastive Learning and Semantic Con...

引用

5th International Conference on Artificial Intelligence and Computer Engineering, ICAICE 2024

作者： Zhang, Chuancai Qu, Dan Du, Liming Yang, Kaiyuan School of Information Systems Engineering University of Information Engineering Zhengzhou China Laboratory for Advanced Computing and Intelligence Engineering Zhengzhou China School of Cyberspace Security Zhengzhou University Zhengzhou China

ISBN: (纸本)9798331528911

In Unsupervised Neural Machine Translation (UNMT) tasks, the lack of extensive parallel corpora makes it challenging for the model to directly optimize the correspondence between the source and target languages. UNMT models primarily rely on learning from monolingual data, making it difficult to fully capture cross-linguistic semantic alignment and consistency during training. Therefore, effectively enhancing semantic alignment between the source and target languages under unsupervised conditions has become a key challenge in the field of UNMT. To address this issue, this study proposes a semantic alignment and consistency enhancement mechanism based on contrastive learning, aiming to improve the model's ability to capture cross-lingual semantic differences by constructing an optimization objective focused on semantic consistency. The method utilizes a pretrained cross-lingual model to generate semantic embeddings for both the source and target languages, then designs a contrastive learning-based loss function that minimizes the semantic distance between correct translations and source sentences while increasing the semantic distance between incorrect translations and source sentences, thereby enhancing the model's performance in semantic consistency. The experiments validate the effectiveness and broad applicability of the semantic consistency optimization method based on contrastive learning, demonstrating great potential and advantages, especially in multilingual tasks of UNMT. The model achieved significant improvements in BLEU scores across various language pairs. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Interest Points Analysis for Internet Forum Based on Long-Short Windows Similarity

引用

Computers, Materials & Continua 2022年第8期72卷 3247-3267页

作者： Xinghai Ju Jicang Lu Xiangyang Luo Gang Zhou Shiyu Wang Shunhang Li Yang Yang State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou450001China School of Computing and Information Systems Singapore Management University188065Singapore College of Computer Science and Big Data Fuzhou UniversityFuzhou350116China

For Internet forum Points of Interest(PoI),existing analysis methods are usually lack of usability analysis under different conditions and ignore the long-term variation,which lead to blindness in method *** address this problem,this paper proposed a PoI variation prediction framework based on similarity analysis between long and short *** on the framework,this paper presented 5 PoI analysis algorithms which can be categorized into 2 types,i.e.,the traditional sequence analysis methods such as autoregressive integrated moving average model(ARIMA),support vector regressor(SVR),and the deep learning methods such as convolutional neural network(CNN),long-short term memory network(LSTM),Transformer(TRM).Specifically,this paper firstly divides observed data into long and short windows,and extracts key words as PoI of each ***,the PoI similarities between long and short windows are calculated for training and ***,series of experiments is conducted based on real Internet forum *** results show that,all the 5 algorithms could predict PoI variations well,which indicate effectiveness of the proposed *** the length of long window is small,traditional methods perform better,and SVR is the *** the contrary,the deep learning methods show superiority,and LSTM performs *** results could provide beneficial references for PoI variation analysis and prediction algorithms selection under different parameter configurations.

关键词： Point of interest(PoI)analysis long and short windows sequential analysis deep learning

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：