AI & 보안 관련 글 정리 (0508)

1. Using deep learning to solve computer security challenges: a survey

 

using Deep Learning techniques to solve computer security challenges

 

security-oriented program analysis, defending return-oriented programming (ROP) attacks, achieving control-flow integrity (CFI), defending network attacks, malware classification, system-event-based anomaly detection, memory forensics, and fuzzing for software security

 

2. A Review of Deep Learning Security and Privacy Defensive Techniques

Deep Learning Private Data Frameworks

Deep Learning Threats and Attacks

Defense Techniques against Security Issues in Deep Learning

 

3. 5 Amazing Applications of Deep Learning in Cybersecurity

https://www.datto.com/blog/5-amazing-applications-of-deep-learning-in-cybersecurity

5 Amazing Applications of Deep Learning in Cybersecurity

4. On the Effectiveness of Machine and Deep Learning for Cyber Security

machine learning techniques applied to the detection of intrusion, malware, and spam

 

 

5. A survey on deep learning for cybersecurity: Progress, challenges, and opportunities

This paper comprehensively presents the promising applications of deep learning, a subfield of AI
based on multiple layers of artificial neural networks, in a wide variety of security tasks. Before critically
and comparatively surveying state-of-the-art solutions from the literature, we discuss the key characteristics of
representative deep learning architectures employed in cybersecurity applications, we introduce the emerging
trends in deep learning, and we provide an overview of necessary resources like a generic framework and
suitable datasets. We identify the limitations of the reviewed works, and we bring forth a vision of the current
challenges of the area, providing valuable insights and good practices for researchers and developers working
on related problems. Finally, we uncover current pain points and outline directions for future research to
address them.

 

6. Emerging challenges and perspectives in Deep Learning model security: A brief survey

In this work, we provide an overview of
some strenuous security risks and concerns that can affect such models. Our focus is on the research challenges
and defense opportunities of the underlying ML framework, when it is devised in specific contexts that can
compromise its effectiveness. Specifically, the survey provides an overview of the following emerging topics:
Model Watermarking, Information Hiding issues and defense opportunities, Adversarial Learning and model
robustness, and Fairness-aware models.

 

7. A Systematic Review on Machine Learning and Deep Learning Models for Electronic Information Security in Mobile Networks