检索结果-内蒙古大学图书馆

2024 IEEE International Conference on Industry 4.0, Artificial Intelligence, and Communications Technology, IAICT 2024

作者： Alamsyah, Samuel Eka Djamhari, Simen Ferdinand Sudjono, Salomo Hendrian Gunawan, Alexander A.S. Setiawan, Karli Eka Data Science Program Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia

ISBN: (纸本)9798350353464

Natural disasters, including earthquakes, cyclones, floods, and wildfires, cause significant environmental damage and have emerged as a major global issue. These events can result in loss of life and disrupt communities and economy. Effective disaster management relies on timely and accurate information. With social media becoming a primary source of real-time information during such events, many images of these disasters are shared, though some may not be accurately labeled. Therefore, the purpose of this study is classifying natural disaster images, mainly cyclone, wildfire, flood, earthquake from social media images. This paper is to conduct a comparative analysis over 4 models namely VGG16, VGG19, EfficientNetB0, and ResNet-50. This paper uses a dataset that contains 4428 human-annotated images including cyclones, earthquakes, floods, and wildfires. Results concluded that ResNet-50 and EfficientNetB0 give the best result with both got over 96% average accuracy. ResNet-50 delivers more stable and more convergent results compared to EfficientNetB0, but it delivers slightly less classification accuracy. VGG16 and VGG19 are both outclassed in terms of accuracy, but better in training time, with VGG19 having a more convergent result. ©2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Open-Vocabulary Calibration for Fine-tuned CLIP 41

Open-Vocabulary Calibration for Fine-tuned CLIP

引用

41st International Conference on Machine Learning, ICML 2024

作者： Wang, Shuoyuan Wang, Jindong Wang, Guoqing Zhang, Bob Zhou, Kaiyang Wei, Hongxin Department of Statistics and Data Science Southern University of Science and Technology Shenzhen China Department of Computer and Information Science University of Macau Taipa China William & Mary WilliamsburgVA United States School of Computer Science and Engineering University of Electronic Science and Technology of China China Department of Computer Science Hong Kong Baptist University Hong Kong

Vision-language models (VLMs) have emerged as formidable tools, showing their strong capability in handling various open-vocabulary tasks in image recognition, text-driven visual content generation, and visual chatbots, to name a few. In recent years, considerable efforts and resources have been devoted to adaptation methods for improving the downstream performance of VLMs, particularly on parameter-efficient fine-tuning methods like prompt learning. However, a crucial aspect that has been largely overlooked is the confidence calibration problem in fine-tuned VLMs, which could greatly reduce reliability when deploying such models in the real world. This paper bridges the gap by systematically investigating the confidence calibration problem in the context of prompt learning and reveals that existing calibration methods are insufficient to address the problem, especially in the open-vocabulary setting. To solve the problem, we present a simple and effective approach called Distance-Aware Calibration (DAC), which is based on scaling the temperature using as guidance the distance between predicted text labels and base classes. The experiments with 7 distinct prompt learning methods applied across 11 diverse downstream datasets demonstrate the effectiveness of DAC, which achieves high efficacy without sacrificing the inference speed. Our code is available at https://***/mlstat-Sustech/CLIP Calibration. Copyright 2024 by the author(s)

关键词： Calibration

来源：评论

学校读者我要写书评

暂无评论

Refined Coreset Selection: Towards Minimal Coreset Size under Model Performance Constraints 41

Refined Coreset Selection: Towards Minimal Coreset Size unde...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Xia, Xiaobo Liu, Jiale Zhang, Shaokun Wu, Qingyun Wei, Hongxin Liu, Tongliang School of Computer Science The University of Sydney Australia College of Information Science and Technology Penn State University United States Department of Statistics and Data Science Southern University of Science and Technology China

Coreset selection is powerful in reducing computational costs and accelerating data processing for deep learning algorithms. It strives to identify a small subset from large-scale data, so that training only on the subset practically performs on par with full data. Practitioners regularly desire to identify the smallest possible coreset in realistic scenes while maintaining comparable model performance, to minimize costs and maximize acceleration. Motivated by this desideratum, for the first time, we pose the problem of refined coreset selection, in which the minimal coreset size under model performance constraints is explored. Moreover, to address this problem, we propose an innovative method, which maintains optimization priority order over the model performance and coreset size, and efficiently optimizes them in the coreset selection procedure. Theoretically, we provide the convergence guarantee of the proposed method. Empirically, extensive experiments confirm its superiority compared with previous strategies, often yielding better model performance with smaller coreset sizes. The implementation is available at https://***/xiaoboxia/LBCS. Copyright 2024 by the author(s)

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Online Learning Experience: An Ensemble Approach for Classifying Student Adaptivity Level 4

Enhancing Online Learning Experience: An Ensemble Approach f...

引用

4th International Conference of science and Information Technology in Smart Administration, ICSINTESA 2024

作者： Mutiara, Belinda Florencia Ariyanto, Gabrielle Felicia Gunawan, Alexander Agung Santoso Setiawan, Karli Eka Bina Nusantara University Data Science Program Computer Science Department School of Computer Science Jakarta11480 Indonesia Bina Nusantara University School of Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350376111

The study investigates the increasing demand of online learning as a means of addressing education issues in the context of the COVID-19 epidemic. Online learning requires several adaptations for teaching methods, learning methodologies, and devices needs. The flexibility of both teachers and students is essential to these adaptations. This research uses dataset collected from a survey about student adaptability level in online education. Carrying out the preprocessing is a challenge as the data used in this research has imbalanced value on the target category. Based on this problem, the aim of this research was created, namely to categorize students' adaptivity levels in online learning and also focuses on finding out approaches that can provide solutions to overcome imbalanced values in the dataset. The model uses ensemble methods - Bagging, Boosting, and Voting with machine learning algorithms. There are two models that stand out, with Soft Voting obtaining the best performance with 9 0% accuracy. © 2024 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

HIV Inhibitory Activity Classification with Morgan Fingerprint and Integrated Graph-Based Deep Learning 12

HIV Inhibitory Activity Classification with Morgan Fingerpri...

引用

12th International Conference on Information and Communication Technology, ICoICT 2024

作者： Verna Siswanto, Jocelyn Angelina Sunarya, Caroline Calista, Agnes Agung Santoso Gunawan, Alexander Eka Setiawan, Karli School of Computer Science Bina Nusantara University Data Science Program Computer Science Department Jakarta11480 Indonesia School of Computer Science Computer Science Department Bina Nusantara University Jakarta11480 Indonesia

ISBN: (纸本)9798350363432

This research offers a new perspective on predicting the activity of the HIV virus from the Drug Therapeutics Program (DTP) Antiviral Screen by using the molecular data represented in SMILES notation. The topic has significance as it focuses on a major global health issue using modern computational approaches and has the potential to uncover new antiviral drug candidates, which could eventually save lives and improve public health outcomes. The study addresses the data imbalance between two classes, active and inactive, and employs the Morgan Fingerprint method for feature extraction, along with the Graph Convolutional Network (GCN) and Graph Attention Network (GAT) as the baseline architectures and the fusion of GCN's and GAT's main features as the proposed architecture. The random oversampling technique is applied to alleviate dataset imbalances. However, even though it improved the training process, the performance of the model flopped when the test set was fed into the model. Combining the main features in GCN and GAT, the proposed model was able to do the classification task more accurately. The attention mechanism from GAT allows the model to focus more on the parts that are more relevant and ignore the irrelevant ones. It managed to outperform the baseline models. Despite a high overall accuracy of 94%, the fusion model exhibits significant disparities in precision, recall, and f1-score metrics, potentially due to class imbalance. Random oversampling led to improved training but compromised model performance on the test set. © 2024 IEEE.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Evaluating Public Opinion on the 2024 Indonesian Presidential Election Candidate: An IndoBERT Approach to Twitter Sentiment Analysis 10

Evaluating Public Opinion on the 2024 Indonesian Presidentia...

引用

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

作者： Tanaja, Rafael Nicholas Widjaya, Andrew Johnny Gunawan, Alexander Agung Santoso Setiawan, Karli Eka School of Computer Science Bina Nusantara University Data Science Program Computer Science Department Jakarta11480 Indonesia School of Computer Science Bina Nusantara University Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350363104

As social media significantly shapes societal norms and ethical paradigms, understanding real-time public sentiment provides valuable insights for political parties to evaluate their candidates, key issues, and campaign strategies. This study aims to gauge each presidential candidate's public sentiment from August 1, 2023, until October 31, 2023, using Twitter data. In this study, the authors would like to conduct an IndoBERT with a Fully Connected Layer to predict the sentiment analysis of Indonesian public sentiment toward the 2024 upcoming presidential election. The paper concluded that the IndoBERT model achieved an accuracy of 84.7%. In conclusion, the findings of this study lay the groundwork for more extensive analyses, enriching our comprehension of the intricate dynamics of public opinion in the political sphere. Furthermore, Sentiment Analysis is dynamic and will change as significant events occur. As we look to the future, integrating cutting-edge language models like IndoBERT with real-time social media data emerges as a transformative force in reshaping political analysis. © 2024 IEEE.

关键词： Tweets

来源：评论

学校读者我要写书评

暂无评论

Breast Cancer Image Classification Obtained Through Dynamic Thermography using Deep Learning 4

Breast Cancer Image Classification Obtained Through Dynamic ...

引用

4th International Conference of science and Information Technology in Smart Administration, ICSINTESA 2024

作者： Timadius, Ethan Daniel Wongso, Reyner Baihaqi, Thirza Gunawan, Alexander Agung Santoso Setiawan, Karli Eka Bina Nusantara University School of Computer Science Data Science Program Computer Science Department Jakarta11480 Indonesia Bina Nusantara University School of Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350376111

This paper addresses the vital task of diagnosing breast cancer early and accurately. It acknowledges the difficulty in detecting minor variations in breast cancer patterns through visual examination alone, particularly when analyzing gray scale images. To overcome this challenge, our research concentrates on using image processing models, particularly Convolutional Neural Networks (CNNs), and enhances them with a refined approach to classification. The research methodology encompasses rigorous data collection, preprocessing, and strategic feature selection to enhance model robustness. To address challenges related to data intricacies, the study incorporates transfer learning, refining the model's training process for improved accuracy. The paper not only seeks to elevate diagnostic precision and minimize errors but also aims to provide healthcare professionals with efficient tools for breast cancer detection. Building upon prior research, the study underscores the potential of deep learning methodologies in advancing breast cancer diagnosis. Furthermore, the fine-tuning process involves adapting a pre-trained model to the intricacies of breast cancer patterns, optimizing its performance. The ultimate goal is to contribute to the development of reliable diagnostic tools that empower medical professionals in the fight against breast cancer. The paper concludes by summarizing key findings. Additionally, the model achieved an accuracy of 84%, highlighting its effectiveness in breast cancer detection. The paper also emphasizes the broader significance of leveraging deep learning for enhanced breast cancer detection. © 2024 IEEE.

关键词： Breast Cancer Convolutional Neural Network Deep Learning Image Classification Thermal Image

来源：评论

学校读者我要写书评

暂无评论

Analyzing Nickel Stock Price Prediction: A Comparative Study of Long Short-Term Memory and Convolutional Neural Network Models

Analyzing Nickel Stock Price Prediction: A Comparative Study...

引用

2024 IEEE International Conference on Industry 4.0, Artificial Intelligence, and Communications Technology, IAICT 2024

作者： Irwanto, Ferry Hafizh Sasongko, Muhammad Farrell Febrian, Stevanus Santoso Gunawan, Alexander Agung Setiawan, Karli Eka Data Science Program Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia

ISBN: (纸本)9798350353464

This research delves to predict PT Vale Indonesia Tbk stock price as an experiment on Indonesian stock using three models: naïve, LSTM, and 1D-CNN. Our analysis emphasizes the importance of matching model architectures to data properties. We compare models' performance with fiveday window for predict one-day prediction output. Interestingly, the single-layer LSTM outperforms the 1D-CNN even with similar hyperparameters, showcasing its strength in capturing long-term temporal dependencies crucial for nickel prices. While the 1D-CNN excels at identifying short-term patterns, its limited receptive field hinders long-term dependence. Recognizing the potential of both models, we encourage exploring hybrid architectures combining LSTM and CNN strengths for further improvement in financial forecasting. The experiment result shows single-layer LSTM outperforms a 1D-CNN with similar settings. ©2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Malicious Query Recognition Using Chosen Machine Learning Techniques

引用

SN computer science 2025年第3期6卷 1-13页

作者： Daramola, Comfort Yetunde Akinpelu, Samson Adebisi Akinyemi, Ezekiel Olajuwon Ajagbe, Sunday Adeola Akinpelu, Gabriel Akinyemi Adigun, Mathew Olusegun Computer Science Department Federal University Oye-Ekiti Nigeria School of Mathematics Statistics & Computer Science University of Kwazulu-Natal Kwazulu-Natal South Africa Department of Computer Science University of Zululand KwaDlangenzwa 3886 South Africa Department of Science Education University of Ilorin Ilorin Nigeria

The successful development of an effective machine learning model for detecting malicious queries is a challenging task that requires domain expertise. Malicious queries are complex and constantly evolving, and their detection is further complicated by variations in database server architectures. To tackle this challenge, domain experts with a deep understanding of attack vectors, techniques, and trends are crucial. Their insights into the patterns, behaviour, and characteristics of malicious queries provide invaluable guidance for designing appropriate features and training the model. This study aims to leverage this domain expertise to propose a robust machine-learning model capable of accurately identifying and categorizing malicious queries. By incorporating expert knowledge and using advanced techniques like machine learning, the model can be optimized to address the dynamic nature of these threats, leading to improved accuracy in real-world scenarios. After the extensive experiment, using 88,213 malicious and normal queries from GitHub and other sources, the Random Forest model yielded the highest accuracy of 98.4% and 97% sensitivity compared to other machine learning models experimented in the study. The performance comparison of the study with existing works indicates that the machine learning model achieved promising results. © The Author(s) 2025.

关键词： Attack Machine learning Malicious query Structure query language Threats

来源：评论

学校读者我要写书评

暂无评论

Viral Melodies: Exploring the Factors Influencing Music Virality in TikTok Engagement 4

Viral Melodies: Exploring the Factors Influencing Music Vira...

引用

4th International Conference of science and Information Technology in Smart Administration, ICSINTESA 2024

作者： Rahardjo, Evelyn Zefanya Alifa, Jasmine Mutia Setiawan, Sharon Zefanya Gunawan, Alexander Agung Santoso Setiawan, Karli Eka Bina Nusantara University Data Science Program Computer Science Department School of Computer Science Jakarta11480 Indonesia Bina Nusantara University School of Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350376111

In the digital era and the evolution of social media platforms like TikTok, understanding the factors influencing content virality has become increasingly crucial. Therefore, this research aims to delve into the musical variables contributing to the popularity of content on TikTok, including aspects such as artist, beat of music, and the number of shares. Focusing on content that successfully garnered over 5, 0 0 0 likes, this study employed a data analysis approach involving the librosa library for music tempo extraction and logistic regression to identify relationships among these variables. The research findings indicate that the most significant variable influencing the virality of music content on TikTok is the number of shares. This highlights that the phenomenon of virality is not solely dependent on musical characteristics like artist or beat but is more influenced by social interactions through video sharing among users. The findings suggest that when someone shares a video with others who may have similar music preferences, the video is likely to receive likes, even if it does not appear on the user's For You Page (FYP). Thus, the virality of content on TikTok can be explained by the existence of social connections among friends that trigger the viral spread of the video. These results provide valuable insights into the mechanisms behind the success of conten/t on this platform. © 2024 IEEE.

关键词： content virality data analysis music selection tiktok user engagement

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：