检索结果-内蒙古大学图书馆

How graph convolutions amplify popularity bias for recommendation?

Frontiers of computer science 2024年第5期18卷 121-132页

作者： Jiajia CHEN Jiancan WU Jiawei CHEN Xin XIN Yong LI Xiangnan HE School of Information Science and Technology University of Science and Technology of ChinaHefei 230026China School of Computer Science and Technology Zhejiang UniversityHangzhou 310058China School of Computer Science and Technology Shandong UniversityQingdao 250100China Department of Electronic Engineering Tsinghua UniversityBeijing 100084China

Graphconvolutional networks(GCNs)have become prevalent in recommender system(RS)due to their superiority in modeling collaborative *** improving the overall accuracy,GCNs unfortunately amplify popularity bias-tail items are less likely to be *** effect prevents the GCN-based RS from making precise and fair recommendations,decreasing the effectiveness of recommender systems in the long *** this paper,we investigate how graph convolutions amplify the popularity bias in *** theoretical analyses,we identify two fundamental factors:(1)with graph convolution(i.e.,neighborhood aggregation),popular items exert larger influence than tail items on neighbor users,making the users move towards popular items in the representation space;(2)after multiple times of graph convolution,popular items would affect more high-order neighbors and become more *** two points make popular items get closer to almost users and thus being recommended more *** rectify this,we propose to estimate the amplified effect of popular nodes on each node's representation,and intervene the effect after each graph ***,we adopt clustering to discover highly-influential nodes and estimate the amplification effect of each node,then remove the effect from the node embeddings at each graph convolution *** method is simple and generic-it can be used in the inference stage to correct existing models rather than training a new model from scratch,and can be applied to various GCN *** demonstrate our method on two representative GCN backbones LightGCN and UltraGCN,verifying its ability in improving the recommendations of tail items without sacrificing the performance of popular *** are open-sourced^(1)).

关键词： recommendation graph convolution networks popularity bias

来源：评论

学校读者我要写书评

暂无评论

A review: data pre-processing techniques used for diabetes prediction 9th

A review: Data pre-processing techniques used for diabetes p...

引用

9th International Conference on computer science and Computational Intelligence, ICCSCI 2024

作者： Isnan, Mahmud Elwirehardja, Gregorius Natanael Pardamean, Bens Bioinformatics and Data Science Research Center Bina Nusantara University Jakarta11480 Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta11480 Indonesia Cmputer Science Department BINUS Graduate Program Computer Science Program Bina Nusantara University Jakarta11480 Indonesia

When processing datasets in diabetes classification, common problems included a large number of missing values, outliers, and dataset imbalance. To deal with those issues, this study analyzed 18 studies on diabetes classification with machine learning algorithms over the past 5 years. This revealed the important role of data pre-processing in creating effective classification models, as it was found that by using different data pre-processing techniques, the same model can provide different performance. The study identified K-Nearest Neighbor (KNN) and support vector machine (SVM) as superior methods for filling in missing values, achieving an accuracy of 98.49% and 94.89%, respectively. These approaches outperformed traditional methods such as median or mean replacement. However, the challenge of imbalanced data sets remains in all studies reviewed. The common evaluation metrics used to evaluate the created models in previous studies included accuracy, precision, specificity, sensitivity/recall, and F1 Score. Overall, this review showed that the role of data pre-processing is no less important than algorithm selection to improve the performance of machine learning models in diabetes classification. © 2024 The Authors.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

CONTESTS: a Framework for Consistency Testing of Span Probabilities in Language Models

CONTESTS: a Framework for Consistency Testing of Span Probab...

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Wagner, Eitan Slavutsky, Yuli Abend, Omri School of Computer Science and Engineering Department of Statistics and Data Science Hebrew University of Jerusalem Israel

ISBN: (纸本)9798891761643

Although language model scores are often treated as probabilities, their reliability as probability estimators has mainly been studied through calibration, overlooking other aspects. In particular, it is unclear whether language models produce the same value for different ways of assigning joint probabilities to word spans. Our work introduces a novel framework, ConTestS (Consistency Testing over Spans), involving statistical tests to assess score consistency across interchangeable completion and conditioning orders. We conduct experiments on post-release real and synthetic data to eliminate training effects. Our findings reveal that both Masked Language Models (MLMs) and autoregressive models exhibit inconsistent predictions, with autoregressive models showing larger discrepancies. Larger MLMs tend to produce more consistent predictions, while autoregressive models show the opposite trend. Moreover, for both model types, prediction entropies offer insights into the true word span likelihood and therefore can aid in selecting optimal decoding strategies. The inconsistencies revealed by our analysis, as well their connection to prediction entropies and differences between model types, can serve as useful guides for future research on addressing these limitations. © 2024 Association for Computational Linguistics.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Edge-IoTDistilBERT: Fine-Tuning DistilBERT for Multi-Class Classification of Network Packet Attacks

Edge-IoTDistilBERT: Fine-Tuning DistilBERT for Multi-Class C...

引用

2025 International Conference on Advancement in data science, E-learning and Information System, ICADEIS 2025

作者： Setiawan, Valentino Soewito, Benfano Binus Graduate Program - Master of Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798331513320

This exponential proliferation of IoT devices is creating an ever-growing demand for efficient cybersecurity solutions in resource-constrained environments. In this study, we propose Edge-IoTDistilBERT, a fine-tuned DistilBERT model, intended for the multi-class classification of network packet attacks within IoT ecosystems. Our proposed model was trained on the Edge-IIoT dataset, representative of all traffic classes and attack variants with high accuracy of up to 99.99%, outperforming state-of-the-art solutions. The new preprocessing pipeline of PCAP to text would transform the network traffic to a textual form allowing the model to learn the pattern in the dataset and making it able to generalize its content. Our fine-tuned model show a robust result on imbalanced dataset with two diffrent splits namely 80-20 and 70-30. The result shown by Edge-IoTDistilBERT place this model as a feasible effective cybersecurity solution for IoT networks, offering a good trade-off between high performance and resource constraints linked to edge devices. © 2025 IEEE.

关键词： Cyber attacks

来源：评论

学校读者我要写书评

暂无评论

Boosting Algorithms in Breast Cancer Classification: Accuracy and Performance Metrics Analysis

Boosting Algorithms in Breast Cancer Classification: Accurac...

引用

International Conference on Green Energy, Computing and Intelligent Technology 2024, GEn-CITy 2024

作者： Oktovianus, Louis Waluya, Kresna B. Surianto, Jeffrey Linardi, Alvin Sunusmo, Maura A. Untoro, Evander B. Edbert, Ivan S. Computer Science Department School of Computer Science Bina Nusantara University Tangerang Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia

ISBN: (纸本)9781837242689

Breast cancer is the most prevalent cancer among women all over the world, so its early detection is of great significance. The traditional methods of detection, while effective, are quite inaccurate. This research evaluates the performance of various boosting algorithms, including AdaBoost, CatBoost, Histogram-Based Gradient Boosting, LightGBM, XGBoost, and Gradient Boosting using the Breast Cancer Wisconsin Original dataset. Our methodology encompassed data exploration, preprocessing, model training, and analysis of results. Among them, the HistGradientBoost model proved to be one of the efficient ones with very high accuracy, precision, recall, and a well-balanced F1-score, besides its excellent AUC score. These results set a great clinical significance for the application of HistGradientBoost in the diagnosis of breast cancer. In addition, this dataset will be expanded, integrated with deep learning techniques, and hyperparameters will be optimized further to enhance the performance of models in the near future. This can provide valuable input to data scientists and medical practitioners in selecting appropriate machine learning models for the diagnosis of breast cancer to arrive at an early diagnosis and prevent poor outcomes. © The Institution of Engineering & Technology 2024.

关键词： Lung cancer

来源：评论

学校读者我要写书评

暂无评论

The Utilization of DistilBERT Features Space for Topics Suggestion in the Context of Learning Content

The Utilization of DistilBERT Features Space for Topics Sugg...

引用

2023 International Workshop on Artificial Intelligence and Image Processing, IWAIIP 2023

作者： Jingga, Kenny Hidayaturrahman, Hidayaturrahman Bina Nusantara University School of Computer Science Computer Science Program Computer Science Department Jakarta Indonesia

ISBN: (纸本)9798350382914

Recommender system is one of the popular topics in artificial intelligence fields as it can widely be used in the *** service provider, e-commerce, e-learning, and many other fields can utilize recommender system to give the personalization for the users. This research will try to use recommender system to provide the recommended topics that are suitable to each learning content. As part of developing the most suitable recommender system, this research will also focus on the data pre-processing,as the data is still raw and contains too much unused information yet. Text vectorization or the embedding process was conducted to the dataset using DistilBERT, pre-trained BERT model. After the vectorization, the recommender system used cosine similarity from the result to discover the largest cosine similarity, which was used to determine the recommendation. Based on the experiment, using cosine similarity could do the recommendation well enough by giving the appropriate topics recommendation based on the content. For example, given the content: 'Chapter 2: Comments Chapter 2 of the book on C programming', the top 5 recommended topics were: 'Objective-C', 'C++', 'MATLAB','VBA', and 'Perl'. Based on the results, it can be considered that the recommender system had performed as expected. However,it still had a lot of areas that could be improved for the future research, especially in the data preprocessing. Other text vectorization models can be considered to be used, such as: BERT Multilingual, RoBERTa, and SpanBERT. Other consideration is the content preparation that will be used as the input for the system. Combination of DistilBERT and cosine similarity as the recommender system can be considered to be implemented for other areas. © 2023 IEEE.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Synonym Replacement Augmentation for Handling data Imbalance in Personality Classification 8

Synonym Replacement Augmentation for Handling Data Imbalance...

引用

8th International Conference on Information Technology, Information Systems and Electrical Engineering, ICITISEE 2024

作者： Pradana, Rilo Chandra Suhartono, Derwin Bina Nusantara University Binus Graduate Program Master of Computer Science Computer Science Department Jakarta11480 Indonesia School of Computer Science Bina Nusantara University Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9798350368970

The Myers-Briggs Type Indicator (MBTI) classification is a widely utilized instrument for personality assessment. However, it frequently encounters challenges due to imbalanced data distributions across personality dimensions. It is paramount to address this issue to enhance the accuracy and reliability of personality predictions. Most current research in this field is focused on data balancing techniques, such as oversampling and undersampling, which have been demonstrated to enhance model performance. Nevertheless, there is a dearth of research exploring text augmentation methods, particularly synonym replacement, for this purpose. This study examines the efficacy of synonym replacement as a data augmentation technique for MBTI classification. Experiments are conducted with varying levels of synonym replacement (10%, 30%, 50%, 70%, and 90%) to assess its impact on model accuracy and F1 scores across the four MBTI dimensions. Our findings indicate that low levels of synonym replacement, particularly at 10% to 50%, can enhance the performance of the model in predicting MBTI dimensions. On the other hand, a higher number of replaced words in the synonym replacement augmentation can harm the model's performance. These observations suggest that synonym replacement can effectively address data imbalance in MBTI classification, although its application must be tailored to specific personality dimensions. © 2024 IEEE.

关键词： data accuracy

来源：评论

学校读者我要写书评

暂无评论

Federated learning-outcome prediction with multi-layer privacy protection

引用

Frontiers of computer science 2024年第6期18卷 205-214页

作者： Yupei ZHANG Yuxin LI Yifei WANG Shuangshuang WEI Yunan XU Xuequn SHANG School of Computer Science Northwestern Polytechnical UniversityXi’an 710129China MIIT Lab of Big Data Storage and Management Xi’an 710129China

Learning-outcome prediction(LOP)is a long-standing and critical problem in educational *** studies have contributed to developing effective models while often suffering from data shortage and low generalization to various institutions due to the privacy-protection *** this end,this study proposes a distributed grade prediction model,dubbed FecMap,by exploiting the federated learning(FL)framework that preserves the private data of local clients and communicates with others through a global generalized *** considers local subspace learning(LSL),which explicitly learns the local features against the global features,and multi-layer privacy protection(MPP),which hierarchically protects the private features,including model-shareable features and not-allowably shared features,to achieve client-specific classifiers of high performance on LOP per *** is then achieved in an iteration manner with all datasets distributed on clients by training a local neural network composed of a global part,a local part,and a classification head in clients and averaging the global parts from clients on the *** evaluate the FecMap model,we collected three higher-educational datasets of student academic records from engineering *** results manifest that FecMap benefits from the proposed LSL and MPP and achieves steady performance on the task of LOP,compared with the state-of-the-art *** study makes a fresh attempt at the use of federated learning in the learning-analytical task,potentially paving the way to facilitating personalized education with privacy protection.

关键词： federated learning local subspace learning hierarchical privacy protection learning outcome prediction privacy-protected representation learning

来源：评论

学校读者我要写书评

暂无评论

Empowering Biomedical Named Entity Recognition Through Multi-Tagger Collaboration 27

Empowering Biomedical Named Entity Recognition Through Multi...

引用

27th European Conference on Artificial Intelligence, ECAI 2024

作者： Zhao, Jin Xie, Jian Zhu, Tinghui Guo, Qian Li, Zhixu Xiao, Yanghua Shanghai Key Laboratory of Data Science School of Computer Science Fudan University China Department of Computer Science and Engineering East China University of Science and Technology China

ISBN: (纸本)9781643685489

Biomedical Named Entity Recognition (BioNER) plays a crucial role in automatically identifying specific categories of entities from biomedical texts. Currently, region-based methods have shown promising performance in BioNER. However, existing paradigms in the region-based methods suffer from inherent limitations, including the generation of negative samples, and the ignorance of token dependencies. To overcome these limitations, we propose a new paradigm, implemented as Token Cascade Tagger (TCT), which combines span identification and category classification. The TCT utilizes category information to enhance the correlation between the heads and tails of entities, effectively reducing the generation of negative samples. Additionally, we introduce a Token Dependency Tagger (TDT) that captures token dependencies within entity spans by identifying the longest span in a sentence. The TDT filters out incorrect spans and further improves the accuracy of span detection obtained from the TCT. Furthermore, we employ a multi-task learning framework to optimize both the TCT and TDT, leading to superior performance in BioNER. Extensive experiments on publicly available biomedical datasets demonstrate our method outperforms the previous state-of-the-art methods, achieving 92.44%, 92.54%, and 81.26% on NCBI-Disease, BC5CDR, and GENIA, respectively, in terms of F1 score. © 2024 The Authors.

关键词： Multi-task learning

来源：评论

学校读者我要写书评

暂无评论

Cryptography Performance Analysis in Intel Core i5-12500H and AMD Ryzen 7 5800H Processors: Implications for Secure Computing 9th

Cryptography Performance Analysis in Intel Core i5-12500H an...

引用

9th International Conference on computer science and Computational Intelligence, ICCSCI 2024

作者： Effendi, Jerry Ong, Albert Sanjaya Wicaksono, Danang Wahyu Computer Science Department School of Computer Science Jakarta11480 Indonesia

Digital security relies heavily on processor-level encryption to ensure the integrity and confidentiality of data. This study compares the cryptographic performance of two modern CPUs, the Intel Core i5-12500H and the AMD Ryzen 7 5800H, focusing on their AES instruction sets. The study benchmarks simulations of cryptographic workloads to evaluate the speed and efficiency of encryption operations supported by the specific instruction sets of these processors. By analyzing the different AES instructions implemented in the microprocessors, the research provides insight into the performance overhead of hardware-accelerated encryption. These findings are critical to understanding the practical applications of cryptographic capabilities in various computing environments and security-sensitive applications. © 2024 The Authors.

关键词： AES-NI Cryptographic workloads Cryptography Digital security Processor-level encryption

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：