Issa Khalil
YOU?
Author Swipe
View article: Explainable AI-Based Intrusion Detection Systems for Industry 5.0 and Adversarial XAI: A Systematic Review
Explainable AI-Based Intrusion Detection Systems for Industry 5.0 and Adversarial XAI: A Systematic Review Open
Industry 5.0 represents a paradigm shift toward human–AI collaboration in manufacturing, incorporating unprecedented volumes of robots, Internet of Things (IoT) devices, Augmented/Virtual Reality (AR/VR) systems, and smart devices. This ex…
View article: PRPO: Paragraph-level Policy Optimization for Vision-Language Deepfake Detection
PRPO: Paragraph-level Policy Optimization for Vision-Language Deepfake Detection Open
The rapid rise of synthetic media has made deepfake detection a critical challenge for online safety and trust. Progress remains constrained by the scarcity of large, high-quality datasets. Although multimodal large language models (LLMs) …
View article: ViGText: Deepfake Image Detection with Vision-Language Model Explanations and Graph Neural Networks
ViGText: Deepfake Image Detection with Vision-Language Model Explanations and Graph Neural Networks Open
The rapid rise of deepfake technology, which produces realistic but fraudulent digital content, threatens the authenticity of media. Traditional deepfake detection approaches often struggle with sophisticated, customized deepfakes, especia…
View article: LLMxCPG: Context-Aware Vulnerability Detection Through Code Property Graph-Guided Large Language Models
LLMxCPG: Context-Aware Vulnerability Detection Through Code Property Graph-Guided Large Language Models Open
Software vulnerabilities present a persistent security challenge, with over 25,000 new vulnerabilities reported in the Common Vulnerabilities and Exposures (CVE) database in 2024 alone. While deep learning based approaches show promise for…
View article: Adaptive Malware Detection using Sequential Feature Selection: A Dueling Double Deep Q-Network (D3QN) Framework for Intelligent Classification
Adaptive Malware Detection using Sequential Feature Selection: A Dueling Double Deep Q-Network (D3QN) Framework for Intelligent Classification Open
Traditional malware detection methods exhibit computational inefficiency due to exhaustive feature extraction requirements, creating accuracy-efficiency trade-offs that limit real-time deployment. We formulate malware classification as a M…
View article: CapsFake: A Multimodal Capsule Network for Detecting Instruction-Guided Deepfakes
CapsFake: A Multimodal Capsule Network for Detecting Instruction-Guided Deepfakes Open
The rapid evolution of deepfake technology, particularly in instruction-guided image editing, threatens the integrity of digital images by enabling subtle, context-aware manipulations. Generated conditionally from real images and textual p…
View article: aiXamine: Simplified LLM Safety and Security
aiXamine: Simplified LLM Safety and Security Open
Evaluating Large Language Models (LLMs) for safety and security remains a complex task, often requiring users to navigate a fragmented landscape of ad hoc benchmarks, datasets, metrics, and reporting formats. To address this challenge, we …
View article: A Client-level Assessment of Collaborative Backdoor Poisoning in Non-IID Federated Learning
A Client-level Assessment of Collaborative Backdoor Poisoning in Non-IID Federated Learning Open
Federated learning (FL) enables collaborative model training using decentralized private data from multiple clients. While FL has shown robustness against poisoning attacks with basic defenses, our research reveals new vulnerabilities stem…
View article: DeBackdoor: A Deductive Framework for Detecting Backdoor Attacks on Deep Models with Limited Data
DeBackdoor: A Deductive Framework for Detecting Backdoor Attacks on Deep Models with Limited Data Open
Backdoor attacks are among the most effective, practical, and stealthy attacks in deep learning. In this paper, we consider a practical scenario where a developer obtains a deep model from a third party and uses it as part of a safety-crit…
View article: StructTransform: A Scalable Attack Surface for Safety-Aligned Large Language Models
StructTransform: A Scalable Attack Surface for Safety-Aligned Large Language Models Open
In this work, we present a series of structure transformation attacks on LLM alignment, where we encode natural language intent using diverse syntax spaces, ranging from simple structure formats and basic query languages (e.g., SQL) to new…
View article: MANTIS: Detection of Zero-Day Malicious Domains Leveraging Low Reputed Hosting Infrastructure
MANTIS: Detection of Zero-Day Malicious Domains Leveraging Low Reputed Hosting Infrastructure Open
Internet miscreants increasingly utilize short-lived disposable domains to launch various attacks. Existing detection mechanisms are either too late to catch such malicious domains due to limited information and their short life spans or u…
View article: STING: A Stealthy Backdoor Attack on GNN-Based Malicious Domain Detection via DNS Perturbations
STING: A Stealthy Backdoor Attack on GNN-Based Malicious Domain Detection via DNS Perturbations Open
Detecting malicious Internet domains is essential for safeguarding against various online threats. The current approach to detecting malicious domains (MDD) employs a graph neural network (GNN) method, which uses DNS logs to construct hete…
View article: PromSec: Prompt Optimization for Secure Generation of Functional Source Code with Large Language Models (LLMs)
PromSec: Prompt Optimization for Secure Generation of Functional Source Code with Large Language Models (LLMs) Open
The capability of generating high-quality source code using large language models (LLMs) reduces software development time and costs. However, they often introduce security vulnerabilities due to training on insecure open-source data. This…
View article: Demo: SGCode: A Flexible Prompt-Optimizing System for Secure Generation of Code
Demo: SGCode: A Flexible Prompt-Optimizing System for Secure Generation of Code Open
This paper introduces SGCode, a flexible prompt-optimizing system to generate\nsecure code with large language models (LLMs). SGCode integrates recent\nprompt-optimization approaches with LLMs in a unified system accessible through\nfront-…
View article: Explainable AI-based Intrusion Detection System for Industry 5.0: An Overview of the Literature, associated Challenges, the existing Solutions, and Potential Research Directions
Explainable AI-based Intrusion Detection System for Industry 5.0: An Overview of the Literature, associated Challenges, the existing Solutions, and Potential Research Directions Open
Industry 5.0, which focuses on human and Artificial Intelligence (AI) collaboration for performing different tasks in manufacturing, involves a higher number of robots, Internet of Things (IoTs) devices and interconnections, Augmented/Virt…
View article: Multi-Instance Adversarial Attack on GNN-Based Malicious Domain Detection
Multi-Instance Adversarial Attack on GNN-Based Malicious Domain Detection Open
Malicious domain detection (MDD) is an open security challenge that aims to detect if an Internet domain is associated with cyber-attacks. Among many approaches to this problem, graph neural networks (GNNs) are deemed highly effective. GNN…
View article: FairDP: Certified Fairness with Differential Privacy
FairDP: Certified Fairness with Differential Privacy Open
This paper introduces FairDP, a novel training mechanism designed to provide group fairness certification for the trained model's decisions, along with a differential privacy (DP) guarantee to protect training data. The key idea of FairDP …
View article: Competent Time Synchronization Mac Protocols to Attain High Performance of Wireless Sensor Networks for Secure Communication
Competent Time Synchronization Mac Protocols to Attain High Performance of Wireless Sensor Networks for Secure Communication Open
Clock synchronization in the Mac layer plays a vital role in wireless sensor network communication that maintains time-based channel sharing and offers a uniform timeframe among different network nodes. Most wireless sensor networks are di…
View article: Ten years after ImageNet: a 360° perspective on artificial intelligence
Ten years after ImageNet: a 360° perspective on artificial intelligence Open
It is 10 years since neural networks made their spectacular comeback. Prompted by this anniversary, we take a holistic perspective on artificial intelligence (AI). Supervised learning for cognitive tasks is effectively solved—provided we h…
View article: Heterogeneous Randomized Response for Differential Privacy in Graph Neural Networks
Heterogeneous Randomized Response for Differential Privacy in Graph Neural Networks Open
Graph neural networks (GNNs) are susceptible to privacy inference attacks (PIAs), given their ability to learn joint representation from features and edges among nodes in graph data. To prevent privacy leakages in GNNs, we propose a novel …
View article: Ten Years after ImageNet: A 360° Perspective on AI
Ten Years after ImageNet: A 360° Perspective on AI Open
It is ten years since neural networks made their spectacular comeback. Prompted by this anniversary, we take a holistic perspective on Artificial Intelligence (AI). Supervised Learning for cognitive tasks is effectively solved - provided w…
View article: An Adaptive Black-box Defense against Trojan Attacks (TrojDef)
An Adaptive Black-box Defense against Trojan Attacks (TrojDef) Open
Trojan backdoor is a poisoning attack against Neural Network (NN) classifiers in which adversaries try to exploit the (highly desirable) model reuse property to implant Trojans into model parameters for backdoor breaches through a poisoned…
View article: Exploration of Enterprise Server Data to Assess Ease of Modeling System Behavior
Exploration of Enterprise Server Data to Assess Ease of Modeling System Behavior Open
Enterprise networks are one of the major targets for cyber attacks due to the vast amount of sensitive and valuable data they contain. A common approach to detecting attacks in the enterprise environment relies on modeling the behavior of …
View article: A Large Scale Study and Classification of VirusTotal Reports on Phishing and Malware URLs
A Large Scale Study and Classification of VirusTotal Reports on Phishing and Malware URLs Open
VirusTotal (VT) provides aggregated threat intelligence on various entities including URLs, IP addresses, and binaries. It is widely used by researchers and practitioners to collect ground truth and evaluate the maliciousness of entities. …
View article: How to Backdoor HyperNetwork in Personalized Federated Learning?
How to Backdoor HyperNetwork in Personalized Federated Learning? Open
This paper explores previously unknown backdoor risks in HyperNet-based personalized federated learning (HyperNetFL) through poisoning attacks. Based upon that, we propose a novel model transferring attack (called HNTroj), i.e., the first …
View article: A Synergetic Attack against Neural Network Classifiers combining Backdoor and Adversarial Examples
A Synergetic Attack against Neural Network Classifiers combining Backdoor and Adversarial Examples Open
In this work, we show how to jointly exploit adversarial perturbation and model poisoning vulnerabilities to practically launch a new stealthy attack, dubbed AdvTrojan. AdvTrojan is stealthy because it can be activated only when: 1) a care…
View article: Time-Window Based Group-Behavior Supported Method for Accurate Detection of Anomalous Users
Time-Window Based Group-Behavior Supported Method for Accurate Detection of Anomalous Users Open
Autoencoder-based anomaly detection methods have been used in identifying anomalous users from large-scale enterprise logs with the assumption that adversarial activities do not follow past habitual patterns. Most existing approaches typic…