检索结果-内蒙古大学图书馆

2nd International Conference on Applied Algorithms, ICAA 2025

作者： Harshitha, Talluri Simhadri, Tanya Jaswanthi, Thadakuluru Rayvanth, N. Singh, Rimjhim Padam Department of Computer Science and Engineering Amrita School of Computing Bengaluru Amrita Vishwa Vidyapeetham Bengaluru India

ISBN: (纸本)9783031845420

With sudden proliferation in evolving Deepfake technology to manipulate media content, identity theft etc., several modern deep learning based solutions are being continuously developed for Deepfake detection, distinguishing the manipulated media. Therefore, the work presented in this paper proposes a novel and real-time MobileNetV3 based Deepfake detection model. With the fine-tuning of MobileNetV3 using additional fully connected layers and self-attention mechanism placed optimally, the proposed model proved effective in concentrating on crucial image regions over relatively lesser important features. The integration of self-attention layers had a direct impact on the model’s capacity to identify the artifacts manipulated during the generation of Deepfakes, thereby, alleviating the detection efficiency. The work also implemented and trained five widely used and recent convolutional models including XceptionNet, InceptionV3, EfficientNetB0, MobileNetV3, and InceptionResNet for best baseline model selection and comparisons against the proposed approach. The proposed model outperformed all other state-of-art models by a margin of 3% with an F-score = 98.46%. The proposed model also has minimalistic memory and time requirements supporting the real-time applicability of the model even on mobile devices. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Accelerating Blockchain-Enabled Federated Learning with Clustered Clients

引用

IEEE Transactions on Big Data 2024年 1-14页

作者： Cui, Laizhong Li, Yinghao Zhou, Yipeng Qu, Youyang Liu, Jiangchuan College of Computer Science and Software Engineering Shenzhen University Shenzhen China School of Computing FSE Macquarie University Macquare Park NSW Australia Data61 CSIRO VIC Australia School of Computing Science Simon Fraser University Burnaby BC Canada

With the rapid development of big data, Federated learning (FL) has found numerous applications, enabling machine learning (ML) on edge devices while preserving privacy. However, FL still faces crucial challenges, such as single point of failure and poisoning attacks, which motivate the integration of blockchain-enabled FL (BeFL). Beyond that, the efficiency issue still limits the further application of BeFL. To address these issues, we propose a novel decentralized framework: Accelerating Blockchain-Enabled Federated Learning with Clustered Clients (ABFLCC), who utilize actual training time for clustering clients to achieve hierarchical FL and solve the single point of failure problem through blockchain. Additionally, the framework clusters edge devices considering their actual training times, which allows for synchronous FL within clusters and asynchronous FL across clusters simultaneously. This approach guarantees that devices with a similar training time have a consistent global model version, improving the stability of the converging process, while the asynchronous learning between clusters enhances the efficiency of convergence. The proposed framework is evaluated through simulations on three real-world public datasets, demonstrating a training efficiency improvement of 30% to 70% in terms of convergence time compared to existing BeFL systems. IEEE

关键词： Big data

来源：评论

学校读者我要写书评

暂无评论

A FRAMEWORK FOR INFERENCE INSPIRED BY HUMAN MEMORY MECHANISMS 12

A FRAMEWORK FOR INFERENCE INSPIRED BY HUMAN MEMORY MECHANISM...

引用

12th International Conference on Learning Representations, ICLR 2024

作者： Zeng, Xiangyu Lin, Jie Hu, Piao Huang, Ruizheng Zhang, Zhicheng Laboratory of Intelligent Collaborative Computing School of Computer Science and Engineering University of Electronic Science and Technology of China Chengdu China

How humans and machines make sense of current inputs for relation reasoning and question-answering while putting the perceived information into context of our past memories, has been a challenging conundrum in cognitive science and artificial intelligence. Inspired by human brain's memory system and cognitive architectures, we propose a PMI framework that consists of perception, memory and inference components. Notably, the memory module comprises working and long-term memory, with the latter endowed with a higher-order structure to retain extensive and complex relational knowledge and experience. Through a differentiable competitive write access, current perceptions update working memory, which is later merged with long-term memory via outer product associations, reducing information conflicts and averting memory overflow. In the inference module, relevant information is retrieved from two separate memory origins and associatively integrated to attain a more comprehensive and precise interpretation of current perceptions. We exploratively apply our PMI to improve prevailing Transformers and CNN models on question-answering tasks like bAbI-20k and Sort-of-CLEVR datasets, as well as detecting equilateral triangles, language modeling and image classification tasks, and in each case, our PMI enhancements consistently outshine their original counterparts significantly. Visualization analyses reveal that relational memory consolidation, along with the interaction and integration of information from diverse memory sources, substantially contributes to the model effectiveness on inference tasks. © 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

关键词： Modeling languages

来源：评论

学校读者我要写书评

暂无评论

Enhanced Bug Detection in Compiler Logs Using Stacking and Explainable AI Techniques 5

Enhanced Bug Detection in Compiler Logs Using Stacking and E...

引用

5th IEEE Global Conference for Advancement in Technology, GCAT 2024

作者： Tripathy, Aayushi Jeeban Rao, Subhiksha Suresh Ashwani, Siddhant Kavitha, C.R. Amrita Vishwa Vidyapeetham Amrita School of Computing Department of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798350376685

The software development process indeed comes up with compiler related bugs, it is critical to identify and fix them correctly to improve software quality. Compiler bugs are very critical to the correctness of software applications;thus, their detection and fixes are important. This paper proposes a new method that uses ensemble learning and explainable AI to improve the identification of compiler bugs. This paper focuses on the ensemble of four classifiers: Random Forest, Support Vector Machine, Neural Network, and Gradient Boosting. Incorporating these models should enhance the predictive accuracy and generalization ability since the individual classifiers have their drawbacks. In order to make the model more interpretable, we use SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) to explain which features affect bug detection decisions. The results reveal the superior performance of the proposed ensemble model which yields better results than each of the individual classifiers in terms of precision, recall, F1-score, and ROC-AUC. In addition to enhancing the bug detection rate with the help of state-of-the-art machine learning and explainable AI, the method even enhances the understanding of compiler log analysis and significantly contributes to the field of software quality assurance. © 2024 IEEE.

关键词： Program compilers

来源：评论

学校读者我要写书评

暂无评论

Multi Head Graph Attention for Drug Response Predicton 3

Multi Head Graph Attention for Drug Response Predicton

引用

3rd International Conference on Smart Data Intelligence, ICSMDI 2023

作者： Selvi Rajendran, P. Sivannarayna, M. Computer Science and Engineering School of Computing Hindustan Institute of Technology and Science Chennai India

ISBN: (纸本)9781665464871

Precision medicine is based on curing diseases based on a patient's genetic profile, lifestyle, and environmental factors. This method improves clinical trial success rates and speed up drug regulatory approval. Predicting tumour vulnerability to specific anti-cancer therapy is critical for the successful implementation of precision medicine. Drug combinations have been shown to be very effective in cancer treatment to lower the drug resistance and improve the therapeutic effectiveness. The experiments carried out in all these therapeutic combinations have become expensive and time-consuming as a result of the increasing number of anti-cancer drugs. Large-scale drug response testing on cancer cell lines might help to understand the way drugs react with cancer cells. This study proposes a multi head graph attention network to perform drug response prediction. © 2023 IEEE.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Comparison of Machine Learning Techniques for Heart Disease Prediction 15

A Comprehensive Comparison of Machine Learning Techniques fo...

引用

15th International Conference on computing Communication and Networking Technologies, ICCCNT 2024

作者： Madhusoodanan, Devika Vismaya, R. Santhosh, Hridyalakshmi Venugopalan, Manju Amrita Vishwa Vidyapeetham Amrita School of Computing Department of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798350370249

Heart disease, affecting millions worldwide, poses an alarming threat to global health due to its high prevalence and significant impact on mortality rates. Despite advances in medical technology, early detection remains crucial for effective intervention and prevention strategies. Machine learning aids in heart disease detection by analyzing patient data to identify patterns and predict risk factors, facilitating early intervention and personalized treatment. In the proposed work, a comprehensive range of classifiers, including AdaBoost, SVM, Naïve Bayes, CatBoost, Random Forest, Perceptron, Decision Tree, XGBoost, kNN, Bagging, Extra Trees, Logistic Regression, Stacking, and Multi-layer Perceptron, were extensively employed to evaluate their effectiveness. Notably, Random Forest exhibited superior performance with an F1 measure of 0.98, endorsing its selection as the primary classifier in our model. © 2024 IEEE.

关键词： Health risks

来源：评论

学校读者我要写书评

暂无评论

Analysis Of The Efficiency of MathBERT and T5 Embeddings in Classification of Students' Responses to Algebraic Questions based on their Correctness and Relevance 5

Analysis Of The Efficiency of MathBERT and T5 Embeddings in ...

引用

5th IEEE Global Conference for Advancement in Technology, GCAT 2024

作者： Manyam, Yashaswini Sundaram, Srinidhi Balakrishnan, Roshni M. Pati, Peeta Basa Amrita Vishwa Vidyapeetham Amrita School of Computing Department of Computer Science & Engineering Bengaluru India

ISBN: (纸本)9798350376685

Automated grading processes helps in saving time for teachers, which can be devoted to more instruction and other essential services. Behavioral analysis of ML models on students' responses to algebraic questions solve problems that exist with traditional manual grading methods thereby easing the manual assessment process. The study involves the application of machine learning classifiers on the 384 dimensional feature extracted using MathBERT and 768 dimensional features extracted using T5 for each response to the algebraic questions in order to analyse their correctness and relevance. The algorithms are trained over the labeled dataset of a wide variety of algebraic answers and provided with the correct and incorrect answers to learn different patterns. The research aims at analyzing the efficiency of MathBERT and T5, which were trained on LaTeX representations of mathematical expressions and massive text, code and symbols data respectively, in generating embeddings that understand the underlying semantics of mathematical expressions. © 2024 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

CTG-Based Fetal Health Prediction: A Comparative Study of Machine Learning Models 15

CTG-Based Fetal Health Prediction: A Comparative Study of Ma...

引用

15th International Conference on computing Communication and Networking Technologies, ICCCNT 2024

作者： Kosuru, Vishnu Vardhan Ramaraju, Sree Virajitha Rajan, Akshay Santhosh, Gouri Anjali, T. Department of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Amritapuri India

ISBN: (纸本)9798350370249

This study explores the use of machine learning approaches for fetal health classification utilizing Cardiotocogram (CTG) data to improve prenatal treatment and maternal-fetal health. The study systematically examines five machine learning algorithms: Random Forest, Gradient Boosting, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Logistic Regression, using an extensive record with 2,126 well-annotated CTG recordings. This work is peculiar since it uses a wide range of feature extraction and preprocessing techniques to ensure high model correctness and dependability. With a 95% accuracy rate, the Random Forest algorithm proven to be the most effective, surpassing the other models by far. This study emphasizes how machine learning, especially ensemble approaches, may effectively predict health problems in fetuses. The study lays the groundwork for future developments in fetal health detection using cutting-edge technologies and machine learning algorithms by emphasizing interpretability and transparency. © 2024 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Comprehensive Music Genre Classification: Exploring Advanced Machine Learning Techniques with the GTZAN Dataset 5

Comprehensive Music Genre Classification: Exploring Advanced...

引用

5th IEEE Global Conference for Advancement in Technology, GCAT 2024

作者： Meghraj, G.K. Nishanth, Y. Kuttan, Akshat Venugopalan, Manju Amrita Vishwa Vidyapeetham Amrita School of Computing Dept of Computer Science and Engineering Bangalore India

ISBN: (纸本)9798350376685

Music is a universal language that comes in a wide variety of genres to suit different interests and moods. In order to achieve this organization, music genre classification - the process of automatically classifying a piece of music according to its auditory content - is essential. A multi-class music genre identification model has been constructed in the proposed study, and ten distinct classifiers, including K-NN, Random Forest, Cat Boost, XG Boost, Decision Tree, etc., have been tested. A popular benchmark dataset for classifying musical genres, the GTZAN dataset offers a wealth of information for developing and assessing classification algorithms. In contrast, CatBoost reports the best outcome, with an F1-score of 0.89. © 2024 IEEE.

关键词： GTZAN Dataset Multi-class Identification Music Genres Performance Measures

来源：评论

学校读者我要写书评

暂无评论

Audio Encryption Using AES and Cellular Automata 15

Audio Encryption Using AES and Cellular Automata

引用

15th International Conference on computing Communication and Networking Technologies, ICCCNT 2024

作者： Reddy, Amireddy Shashank Achar, Dishan Narsimhachari Mol, Madithilveedu Sheethal Panda, Niharika Amrita Vishwa Vidyapeetham Amrita School of Computing Department of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798350370249

Securing digital audio data during any transmission or storage is essential for ensuring privacy and confidentiality. Traditional encryption techniques like Advanced Encryption Standard (AES) offer high security, but they might be vulnerable to attacks that exploit encrypted data patterns. Cellular automata, on the other hand, exhibit complex and challenging behavior, making them attractive for introducing additional layers of security in encryption schemes. However, integrating these techniques effectively remains a challenge. This paper addresses this gap by proposing a novel audio encryption system that combines the features of AES and cellular automata. A pseudo-random number sequence produced by Rule 30, a chaotic cellular automaton rule, is XOR-ed with the AES-encrypted audio data. This disrupts potential pattern recognition attempts, significantly enhancing the overall security of the encryption process. The system operates on WAV format audio files and ensures the confidentiality and integrity of the data by preventing unauthorized access and manipulation. We evaluate the effectiveness of our approach by comparing the characteristics of original and encrypted audio files. The results demonstrate significant alterations in the audio data, rendering it unintelligible without the decryption key. Additionally, the system utilizes the well-established AES algorithm, maintaining efficiency during encryption and decryption. This system defeats other methods in terms of security, complexity, and computational efficiency making this a reliable and efficient method for securing audio data transmission and storage where audio privacy is essential. © 2024 IEEE.

关键词： Advanced Encryption Standard (AES) Audio Encryption Cellular Automata Cryptography Digital Security Rule 30

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：