Evaluating Query Efficiency and Accuracy of Transfer Learning-based Model Extraction Attack in Federated Learning

2025/05/25 by Sayyed Farid Ahamed, Sandip Roy, Ahamed, Sayyed Farid +15
Computer Science · #Artificial Intelligence (cs.AI) #Confidentiality #Cryptography and Security (cs.CR) #D.4.6 #Data modeling #FOS: Computer and information sciences #Federated learning #Fidelity #High fidelity #I.2.6 #Machine Learning (cs.LG) #Point (geometry) #Privacy-Preserving Technologies in Data #Replicate #Transfer of learning #Vulnerability (computing)

paper · pdf · doi:10.48550/arxiv.2505.23791

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2025/05/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05

Abstract

Federated Learning (FL) is a collaborative learning framework designed to protect client data, yet it remains highly vulnerable to Intellectual Property (IP) threats. Model extraction (ME) attacks pose a significant risk to Machine Learning as a Service (MLaaS) platforms, enabling attackers to replicate confidential models by querying black-box (without internal insight) APIs. Despite FL's privacy-preserving goals, its distributed nature makes it particularly susceptible to such attacks. This paper examines the vulnerability of FL-based victim models to two types of model extraction attacks. For various federated clients built under the NVFlare platform, we implemented ME attacks across two deep learning architectures and three image datasets. We evaluate the proposed ME attack performance using various metrics, including accuracy, fidelity, and KL divergence. The experiments show that for different FL clients, the accuracy and fidelity of the extracted model are closely related to the size of the attack query set. Additionally, we explore a transfer learning based approach where pretrained models serve as the starting point for the extraction process. The results indicate that the accuracy and fidelity of the fine-tuned pretrained extraction models are notably higher, particularly with smaller query sets, highlighting potential advantages for attackers.

Citations

• Model Extraction Attacks Revisited
• MIA-BAD: An Approach for Enhancing Membership Inference Attack and its Mitigation with Federated Learning
• Model Extraction Attacks on Split Federated Learning
• Anomaly Detection via Federated Learning
• Source Inference Attacks in Federated Learning
• De-Pois: An Attack-Agnostic Defense against Data Poisoning Attacks
• Data-Free Model Extraction
• On Safeguarding Privacy and Security in the Framework of Federated Learning
• High Accuracy and High Fidelity Extraction of Neural Networks
• Model Extraction Warning in MLaaS Paradigm
• Stealing Machine Learning Models via Prediction APIs

Related

• Model Extraction Attacks on Split Federated Learning
• Model Extraction Attacks Revisited
• Local Model Reconstruction Attacks in Federated Learning and their Uses
• Exploring Connections Between Active Learning and Model Extraction
• Federated Learning Attacks Revisited: A Critical Discussion of Gaps, Assumptions, and Evaluation Setups