#258 サイバーarxiv (2025-09-03)

本日紹介した論文の一覧 SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models http://arxiv.org/abs/2509.03487v1 Evaluating Diverse Feature Extraction Techniques of Multifaceted IoT Malware Analysis: A Survey http://arxiv.org/abs/2509.03442v1 Federated Learning: An approach with Hybrid Homomorphic Encryption http://arxiv.org/abs/2509.03427v1 Exposing Privacy Risks in Anonymizing Clinical Data: Combinatorial Refinement Attacks on k-Anonymity Without Auxiliary Information http://arxiv.org/abs/2509.03350v1 VulnRepairEval: An Exploit-Based Evaluation Framework for Assessing Large Language Model Vulnerability Repair Capabilities http://arxiv.org/abs/2509.03331v1 Evaluating Security Properties in the Execution of Quantum Circuits http://arxiv.org/abs/2509.03306v1 A Comprehensive Guide to Differential Privacy: From Theory to User Expectations http://arxiv.org/abs/2509.03294v1 Kangaroo: A Private and Amortized Inference Framework over WAN for Large-Scale Decision Tree Evaluation http://arxiv.org/abs/2509.03123v1 PromptCOS: Towards System Prompt Copyright Auditing for LLMs via Content-level Output Similarity http://arxiv.org/abs/2509.03117v1 Compressed verification for post-quantum signatures with long-term public keys http://arxiv.org/abs/2509.03098v1 EverTracer: Hunting Stolen Large Language Models via Stealthy and Robust Probabilistic Fingerprint http://arxiv.org/abs/2509.03058v1 TraceLLM: Security Diagnosis Through Traces and Smart Contracts in Ethereum http://arxiv.org/abs/2509.03037v1 Efficient Privacy-Preserving Recommendation on Sparse Data using Fully Homomorphic Encryption http://arxiv.org/abs/2509.03024v1 Closing the Visibility Gap: A Monitoring Framework for Verifiable Open RAN Operations http://arxiv.org/abs/2509.03000v1 Integrating Generative AI into Cybersecurity Education: A Study of OCR and Multimodal LLM-assisted Instruction http://arxiv.org/abs/2509.02998v1 なお、ポッドキャスト内で紹介する内容は、各論文の概要を日本語で解説したもので、論文概要の著作権は論文著者に帰属します。]]

There are no comments yet.

Please log in to write the first comment.