E2EGI: End-to-End Gradient Inversion in Federated Learning

Article Engineering, Multidisciplinary

EPPDA: An Efficient Privacy-Preserving Data Aggregation Federated Learning Scheme

Jingcheng Song et al.

Summary: Federated learning is a privacy-aware machine learning method that trains models on users' devices and aggregates the updates on a server. However, it is vulnerable to reverse attacks, where adversaries can analyze user-uploaded models to obtain users' data. This paper proposes an efficient privacy-preserving data aggregation mechanism, based on secret sharing, to resist reverse attacks and securely aggregate users' trained models. The mechanism also has efficient fault tolerance and protects users' privacy without compromising efficiency.

IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Leveraging transferability and improved beam search in textual adversarial attacks

Bin Zhu et al.

Summary: Adversarial attacks in NLP are difficult to defend due to the discrete and abstract nature of human languages. Previous studies have used different word replacement strategies to generate semantically preserved adversarial texts. However, these query-based methods have limited exploration of the search space. This study proposes an improved beam search algorithm and utilizes the transferable vulnerability between models to select vulnerable candidate words. Experimental results show that our method outperforms three advanced attacking methods under black-box settings.

NEUROCOMPUTING (2022)

添加到收藏夹

Article Computer Science, Cybernetics

InfusedHeart: A Novel Knowledge-Infused Learning Framework for Diagnosis of Cardiovascular Events

Sharnil Pandya et al.

Summary: In this study, a customized dataset and a benchmark dataset were used for analyzing and evaluating heartbeat acoustic events. An audio processing library was used to extract more information, and various machine learning and deep learning methods were applied for classification. The proposed InfusedHeart Framework outperformed other methods with high accuracy, as shown by the experimental results.

IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS (2022)

添加到收藏夹

Article Telecommunications

IEPSBP: A Cost-Efficient Image Encryption Algorithm Based on Parallel Chaotic System for Green IoT

Zhaoquan Gu et al.

Summary: With the development of IoT, data privacy and security threats have become emerging challenges in green IoT image transmission. To address the limited computing power of devices, a parallel chaotic system named PSBP is proposed, along with a cost-efficient image encryption algorithm IEPSBP based on PSBP.

IEEE TRANSACTIONS ON GREEN COMMUNICATIONS AND NETWORKING (2022)

添加到收藏夹

Article Computer Science, Information Systems

CorrAUC: A Malicious Bot-IoT Traffic Detection Method in IoT Network Using Machine-Learning Techniques

Muhammad Shafiq et al.

Summary: Researchers have proposed a new feature selection method and algorithm to accurately detect malicious traffic in IoT networks. By integrating TOPSIS and Shannon entropy methods to validate the selected features for malicious traffic identification in IoT networks, the experimental results have shown that this method is efficient and can achieve over 96% accuracy on average.

IEEE INTERNET OF THINGS JOURNAL (2021)

添加到收藏夹

Article Computer Science, Information Systems

Variation-Aware Federated Learning With Multi-Source Decentralized Medical Image Data

Zengqiang Yan et al.

Summary: The paper introduces a variation-aware federated learning (VAFL) framework to address the cross-client variation problem in medical image data by minimizing variations among clients while preserving privacy, used for automated classification of clinically significant prostate cancer.

IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS (2021)

添加到收藏夹

Article Engineering, Multidisciplinary

Gradient Shielding: Towards Understanding Vulnerability of Deep Neural Networks

Zhaoquan Gu et al.

Summary: This paper proposes a new method of attacking deep neural networks by gradient shielding, which ignores insensitive information in images and fools image classification DNNs by adding perturbations. Experimental results demonstrate the effectiveness of this method, leading to a first place win in a competition.

IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING (2021)

添加到收藏夹

Proceedings Paper Computer Science, Hardware & Architecture

Label Leakage from Gradients in Distributed Machine Learning

Aidmar Wainakh et al.

Summary: In this research, the privacy risk of disclosing users' training data labels through shared gradients in distributed machine learning is highlighted, and the efficacy of the proposed LLG algorithm is demonstrated through empirical analysis on two datasets, showing high accuracy in extracting labels in various scenarios.

2021 IEEE 18TH ANNUAL CONSUMER COMMUNICATIONS & NETWORKING CONFERENCE (CCNC) (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Image fusion practice to improve the ischemic-stroke-lesion detection for efficient clinical decision making

D. Jude Hemanth et al.

Summary: The research proposes a new diagnostic method through MRI-based IS assessment and confirms that fusing multi-modal brain MRI slices can improve lesion extraction accuracy, which is crucial for clinical-level disease diagnosis.

EVOLUTIONARY INTELLIGENCE (2021)

添加到收藏夹

Article Computer Science, Theory & Methods