检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Lu, Shao-Chien Yeh, Chen-Chen Cho, Hui-Lin Hsu, Chun-Chieh Hsu, Tsai-Ling Wu, Cheng-Han Shih, Timothy K. Lin, Yu-Cheng Department of Computer Science and Engineering Yuan Ze University Taoyuan Taiwan International Bachelor Program in Informatics Yuan Ze University Taoyuan Taiwan Department of Computer Science and Information Engineering National Central University Taoyuan Taiwan

The field of music generation using Large Language Models (LLMs) is evolving rapidly, yet existing music notation systems, such as MIDI, ABC Notation, and MusicXML, remain too complex for effective fine-tuning of LLMs. These formats are difficult for both machines and humans to interpret due to their variability and intricate structure. To address these challenges, we introduce YNote, a simplified music notation system that uses only four characters to represent a note and its pitch. YNote’s fixed format ensures consistency, making it easy to read and more suitable for fine-tuning LLMs. In our experiments, we fine-tuned GPT-2 (124M) on a YNote-encoded dataset and achieved BLEU and ROUGE scores of 0.883 and 0.766, respectively. With just two notes as prompts, the model was able to generate coherent and stylistically relevant music. We believe YNote offers a practical alternative to existing music notations for machine learning applications and has the potential to significantly enhance the quality of music generation using LLMs. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Transfer Learning for Keypoint Detection in Low-Resolution Thermal TUG Test Images

arXiv

引用

arXiv 2025年

作者： Chen, Wei-Lun Hsieh, Chia-Yeh Kao, Yu-Hsiang Liu, Kai-Chun Peng, Sheng-Yu Tsao, Yu Research Center for Information Technology Innovation Academic Sinica Taiwan Graduate Institute of Electrical Engineering National Taiwan University Taiwan Bachelor’s Program in Medical Informatics and Innovative Applications Fu Jen Catholic University Taiwan Department of Electrical Engineering National Taiwan University Taiwan College of Information and Computer Sciences University of Massachusetts AmherstMA01003 United States Department of Electrical Engineering National Taiwan University of Science of Technology Taiwan

This study presents a novel approach to human keypoint detection in low-resolution thermal images using transfer learning techniques. We introduce the first application of the Timed Up and Go (TUG) test in thermal image computer vision, establishing a new paradigm for mobility assessment. Our method leverages a MobileNetV3-Small encoder and a ViTPose decoder, trained using a composite loss function that balances latent representation alignment and heatmap accuracy. The model was evaluated using the Object Keypoint Similarity (OKS) metric from the COCO Keypoint Detection Challenge. The proposed model achieves better performance with AP, AP50, and AP75 scores of 0.861, 0.942, and 0.887 respectively, outperforming traditional supervised learning approaches like Mask R-CNN and ViTPose-Base. Moreover, our model demonstrates superior computational efficiency in terms of parameter count and FLOPS. This research lays a solid foundation for future clinical applications of thermal imaging in mobility assessment and rehabilitation monitoring. Copyright © 2025, The Authors. All rights reserved.

关键词： Thermography (imaging)

来源：评论

学校读者我要写书评

暂无评论

Entropy and information gain analysis on low cost BCI for motorbike users to prevent accident 4

Entropy and information gain analysis on low cost BCI for mo...

引用

4th International Conference on Information Technology, Information Systems and Electrical engineering, ICITISEE 2019

作者： Awangga, Rolly Maulana Pane, Syafrial Fachri Majesty, Dinda Setiawan, Asep Asyhari, Moch Yusuf Applied Bachelor Program of Informatics Engineering Politeknik Pos Indonesia Bandung Indonesia Universitas Islam Indonesia Magister Program of Informatics Engineering Yogyakarta Indonesia

ISBN: (纸本)9781728151182

The phenomenon of mothers giving the wrong vehicle signal lights to turn left and right reminds the effect of giving the right signal lights to other road users. Weak human concentration while driving can cause errors when giving vehicle signal lights or not giving vehicle signal lights. Mistakes when giving a signal light a vehicle has a risk of an accident. The driver's brain wave data will show the results of the rider's brain thinking when turning. This study aims to analyze the concentration of the driver when driving. This research will obtain driver brain wave data when driving and analyze human brain waves while riding a motorcycle, especially when turning left and right, to reduce the risk of accidents while driving. The driver will use brain waves tool while driving, and the computer will detect the driver's brain waves, and then send brainwave data to the computer. Driver brain waves are analyzed using the formula entropy and gain. The results of the driver's brain waves become a decision whether the driver is allowed to drive on the highway based on the detected waves (alpha, beta, gamma, theta, and delta) to reduce the chance of an accident. © 2019 IEEE.

关键词： Entropy

来源：评论

学校读者我要写书评

暂无评论

Performance Evaluation of Oversampling Methods on Deep Learning-Based Skin Cancer Classification

Performance Evaluation of Oversampling Methods on Deep Learn...

引用

2023 International Seminar on Application for Technology of Information and Communication, iSemantic 2023

作者： Supriyanto, Catur Salam, Abu Zeniarja, Junta Wijaya, Adi Universitas Dian Nuswantoro Bachelor Program in Informatics Engineering Faculty of Computer Science Semarang Indonesia Universitas Indonesia Maju Department of Health Information Management Jakarta Indonesia

ISBN: (纸本)9798350339215

The field of dermatology faces considerable challenges when it comes to early detection of skin cancer. Our study focused on using different datasets, including original data, augmented data, and SMOTE oversampled data, to identify skin cancer. Our dataset consisted of images of skin lesions from the MNIST Skin Cancer dataset (HAM 10000), including samples of both cancerous and benign cases in the dataset. We employed data augmentation to expand the dataset's size and increase the diversity of skin lesion features. Furthermore, to tackle class imbalance in the dataset, we applied the SMOTE oversampling technique to generate synthetic samples for the under-represented group. With the original, augmented, and SMOTE oversampled datasets, we trained a Convolutional Neural Network (CNN) model. The performance of the model was evaluated using accuracy, recall, precision, and F1-score. The comparison between the results obtained from the original data, augmented data, and SMOTE oversampling data clearly revealed distinctions in performance. Our findings clearly demonstrate that employing data augmentation and SMOTE oversampling can significantly enhance the efficacy of skin cancer detection. © 2023 IEEE.

关键词： Neural network models

来源：评论

学校读者我要写书评

暂无评论

Comparison Of Multinomial Naive Bayes Algorithm And Logistic Regression For Intent Classification In Chatbot

Comparison Of Multinomial Naive Bayes Algorithm And Logistic...

引用

International Conference on Applied engineering (ICAE)

作者： Muhammad Yusril Helmi Setyawan Rolly Maulana Awangga Safif Rafi Efendi Applied Bachelor Program of Informatics Engineering Politeknik Pos Indonesia Bandung Indonesia

ISBN: (纸本)9781538680674

Chatbot is software that communicates using natural language. chatbots such as machine conversation systems, Chatterbot, virtual agents, and dialogue systems. This software enables to simulate human conversations. In this research, the chatbot system that will be created must be able to understand the natural language of what is entered by the user, and the chatbot will answer according to what the user is expecting. The researcher proposes a classification method to identify intent rather than user input or called intent classification on the chatbot system; the researcher also wants to know the level of accuracy, precision, and recall on the evaluation results of both methods. The classification method applied in this research is the Naive Bayes method and compared with the Logistic Regression method to determine the class intention. The evaluation results show the level of accuracy precision and recall in the Logistic Regression model is higher than the Naive Bayes model.

关键词： Logistics Training data Bayes methods Mathematical model Data models Classification algorithms Natural languages

来源：评论

学校读者我要写书评

暂无评论

Multi-Speaker Voice Cloning in Indonesian Using Sequence to Sequence Speech Synthesis

SSRN

引用

SSRN 2022年

作者： Awangga, Rolly Maulana Majesty, Dinda Applied Bachelor Program of Informatics Engineering Politeknik Pos Indonesia Bandung40151 Indonesia

Multi-Speaker Voice Cloning is a system that can produce various voices based on voice samples and text input. Voice cloning can be an alternative to get audio data quickly without recording, which requires a lot of time, cost, and effort. Sequence to Sequence Speech Synthesis produces sounds that are naturally similar to speech equivalent to the human voice. Sequence to Sequence Speech Synthesis is a technology for generating speech directly from the text, consisting of attention-based predictive network features. Multi-Speaker Voice Clone consists of speaker encoder, encoder-decoder, and vocoder models. Tacotron-2 as model encoder-decoder and WaveNet as a vocoder. The speaker encoder will produce a speaker embedding to recognize the sound from the speaker. Tacotron-2 will embed text into the mel scale spectrogram. WaveNet will synthesize the time domain waveform from the mel scale spectrogram and generate sound predictions based on the samples. This model uses a multi-speaker voice dataset in Indonesian. Combining the three models produces a value of 3.28 for speech naturalness and 1.94 for speech similarity. © 2022, The Authors. All rights reserved.

关键词： Cloning

来源：评论

学校读者我要写书评

暂无评论

Entropy and Information Gain Analysis on Low Cost BCI for Motorbike Users to Prevent Accident

Entropy and Information Gain Analysis on Low Cost BCI for Mo...

引用

International Conference on Information Technology, Information Systems and Electrical engineering (ICITISEE)

作者： Rolly Maulana Awangga Syafrial Fachri Pane Dinda Majesty Asep Setiawan Moch Yusuf Asyhari Applied Bachelor Program of Informatics Engineering Politeknik Pos Indonesia Bandung Indonesia Magister Program of Informatics Engineering Universitas Islam Indonesia Yogyakarta Indonesia

ISBN: (数字)9781728151182

ISBN: (纸本)9781728151199

关键词： Electroencephalography Entropy Turning Tools Calibration Motorcycles

来源：评论

学校读者我要写书评

暂无评论

A VAE-Based Data Augmentation Method for Fall Detection System using Wearable Sensor 10

A VAE-Based Data Augmentation Method for Fall Detection Syst...

引用

10th International Conference on Applied System Innovation, ICASI 2024

作者： Tu, Yu-Chen Wang, Hsuan-Chih Liu, Chien-Pin Hsieh, Chia-Yeh Chan, Chia-Tai National Yang Ming Chiao Tung University Department of Biomedical Engineering Taipei City112 Taiwan Bachelor's Program in Medical Informatics and Innovative Applications Fu Jen Catholic University New Taipei City242062 Taiwan

ISBN: (纸本)9798350394924

Fall incidents among the elderly represent a significant global concern, often resulting in physical injuries and psychological distress. It is crucial to develop reliable fall detection systems which are capable of identifying fall events immediately and triggering alerts for assistance. However, real-world fall occurrences are infrequent, leading to a highly imbalanced class situation. Training a model with imbalanced datasets may result in biased models with poor performance in fall detection. To address this challenge, various techniques such as data transformation and Synthetic Minority Oversampling Technique (SMOTE) have been proposed. However, these methods are constrained by issues such as limitations in input data size or sensitivity to outliers. Compared to other methods, variational autoencoder (VAE) can generate data with a similar probability distribution to the original input data while constraining the latent representation in a Gaussian distribution. This study proposes a VAE-based data augmentation method for wearable-based fall detection system. The proposed method is validated on the FallAllD public dataset, achieving a F-score of 99.46%. The performance has been increased by 2.21%. The results demonstrate the effectiveness of VAE-based data augmentation technique in enhancing fall detection systems and its superior performance compared with other traditional data augmentation methods. © 2024 IEEE.

关键词： Wearable sensors

来源：评论

学校读者我要写书评

暂无评论

Prefallkd: Pre-Impact Fall Detection Via CNN-ViT Knowledge Distillation 48

Prefallkd: Pre-Impact Fall Detection Via CNN-ViT Knowledge D...

引用

48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

作者： Chi, Tin-Han Liu, Kai-Chun Hsieh, Chia-Yeh Tsao, Yu Chan, Chia-Tai National Yang Ming Chiao Tung University Department of Biomedical Engineering Taiwan Academia Sinica Research Center for Information Technology Innovation Taiwan Fu Jen Catholic University Bachelor's Program in Medical Informatics and Innovative Applications Taiwan

ISBN: (纸本)9781728163277

Fall accidents are critical issues in an aging and aged society. Recently, many researchers developed "pre-impact fall detection systems"using deep learning to support wearable-based fall protection systems for preventing severe injuries. However, most works only employed simple neural network models instead of complex models considering the usability in resource-constrained mobile devices and strict latency requirements. In this work, we propose a novel pre-impact fall detection via CNN-ViT knowledge distillation, namely PreFallKD, to strike a balance between detection performance and computational complexity. The proposed PreFallKD transfers the detection knowledge from the pre-trained teacher model (vision transformer) to the student model (lightweight convolutional neural networks). Additionally, we apply data augmentation techniques to tackle issues of data imbalance. We conduct the experiment on the KFall public dataset and compare PreFallKD with other state-of-the-art models. The experiment results show that PreFallKD could boost the student model during the testing phase and achieves reliable F1-score (92.66%) and lead time (551.3 ms). © 2023 IEEE.

关键词： Inertial measurement units Knowledge distillation Pre-impact fall detection Vision transformer Wearable sensors

来源：评论

学校读者我要写书评

暂无评论

An Analysis of Customer Agrotourism Resort Behaviour based on RFM and Mean Shift Clustering

An Analysis of Customer Agrotourism Resort Behaviour based o...

引用

International Conference on Applied engineering (ICAE)

作者： Muhammad Ruslan Maulani Syafrial Fachri Pane Rolly Maulana Awangga Diana Asri Wijayanti Wahyu Caesarendra Diploma Program of Informatics Engineering Politeknik Pos Indonesia Bandung Indonesia Applied Bachelor Program of Informatics Engineering Politeknik Pos Indonesia Bandung Indonesia School of Mechanical and Aerospace Engineering Nanyang Technological University Singapore

ISBN: (纸本)9781538680674

This paper presents an analysis study of customer reservation behavior in Agrotourism N8 based on recency, frequency and monetary (RFM) method. The historical data from January to December 2016 were collected from Rancabali tour and used for RFM analysis. Prior to the RFM analysis, the data were grouped into frequency and monetary using mean shift clustering. This analysis will provide the guideline to Agrotourism N8 to predict the customer needs in the future based on historical data. The RFM method shows that Rancabali tour always crowded with visitors until the end of each month. The highest number of visitors occurred in January, July, and December. Moreover, The highest revenue earned in December. The mean shift shows the income-based cluster by the frequency of customer.

关键词： Companies Market research Bandwidth informatics Kernel Industries

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：