IDUNN Project

𝗘𝗻𝗵𝗮𝗻𝗰𝗶𝗻𝗴 𝟱𝗚 𝗖𝘆𝗯𝗲𝗿𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘄𝗶𝘁𝗵 𝗙𝗲𝗱𝗲𝗿𝗮𝘁𝗲𝗱 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗮𝗻𝗱 𝗟𝗦𝗧𝗠 𝗡𝗲𝘁𝘄𝗼𝗿𝗸𝘀 As 5G continues to revolutionize mobile communication, it also introduces new challenges for safeguarding networks from sophisticated cyber threats. This latest research, authored by Saeid Sheikhi and Panos Kostakos, introduces an innovative solution: a combination of Federated Learning and Long Short-Term Memory (LSTM) networks to enhance cyber threat detection within 5G infrastructures. Key highlights of this study: 📡 Novel FL-LSTM approach to identify cyberattacks targeting the GPRS Tunneling Protocol (GTP) in 5G. 🛡️ Enhanced security and privacy with decentralized learning. 📊 Effective detection of Distributed PFCP and IP Spoofing attacks within a real-world 5G testbed. This research showcases how Federated Learning can preserve privacy and improve security in 5G and future networks. Stay ahead of the curve as 5G expands globally, ensuring robust and resilient cybersecurity measures. #5G #Cybersecurity #FederatedLearning #AI #LSTM #Research #NetworkSecurity #Innovation #Tech https://lnkd.in/dtaXgsGm

https://lnkd.in/eKkHU7aM

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗖𝘆𝗧𝗥𝗜: 𝗘𝗻𝗵𝗮𝗻𝗰𝗶𝗻𝗴 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗠𝗲𝗮𝘀𝘂𝗿𝗲𝘀 𝗔𝗴𝗮𝗶𝗻𝘀𝘁 𝗘𝗺𝗲𝗿𝗴𝗶𝗻𝗴 𝗖𝘆𝗯𝗲𝗿 𝗧𝗵𝗿𝗲𝗮𝘁𝘀 We are excited to present our latest research on #CyTRI, a comprehensive cybersecurity solution designed to tackle the growing vulnerabilities in industrial control systems (#ICS) and operational technology (OT) environments. Developed by experts at OFFIS – Institute for Information Technology, CyTRI offers a robust framework to enhance system resilience against both internal and external cyber threats. 𝘈𝘣𝘰𝘶𝘵 𝘊𝘺𝘛𝘙𝘐: CyTRI, or the Cyber Threat & Risk Intelligence template, is a cutting-edge solution that models threats and hypothesizes potential attack scenarios to analyze the activities of adversaries who may gain access to the system. This proactive approach ensures that security measures are in place to mitigate risks effectively. 𝘒𝘦𝘺 𝘍𝘦𝘢𝘵𝘶𝘳𝘦𝘴: Threat Modeling: Analyzes and predicts potential cyber-attacks to preemptively secure ICS and OT environments. Compliance with IEC 62443: Adheres to international standards to ensure robust security-by-design for industrial control systems. Security-by-Design: Implements security measures from the ground up, making systems more resilient to sophisticated cyber threats. 𝘈𝘶𝘵𝘩𝘰𝘳𝘴: Mana Azamat, OFFIS - Institute for Information Technology Dr. Oliver Werth, OFFIS - Institute for Information Technology Mathias Uslar, OFFIS - Institute for Information Technology 𝘗𝘶𝘣𝘭𝘪𝘤𝘢𝘵𝘪𝘰𝘯 𝘛𝘺𝑝𝘦: 𝘊𝘰𝘮𝑝𝘭𝘦𝘵𝘦 𝘗𝘢𝑝𝘦𝘳 This research is essential for those involved in securing complex industrial systems and looking to adopt proactive cybersecurity measures. Stay tuned for more updates and insights from our research team at OFFIS. If you have any questions or would like to learn more, feel free to reach out! #CyberSecurity #IndustrialControlSystems #OperationalTechnology #ThreatModeling #IEC62443 #SecurityByDesign #Research #OFFIS https://lnkd.in/d8dGir7j

𝗘𝘅𝗰𝗶𝘁𝗶𝗻𝗴 𝗧e𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗰𝗮𝗹 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗜𝗗𝗨𝗡𝗡 𝗣𝗿𝗼𝗷𝗲𝗰𝘁! We are thrilled to introduce our latest advancements in cybersecurity: AMORA, HEIMDAL, THOR, ODIN, and FRIGG. These innovative tools are designed to tackle the ever-evolving challenges in the digital landscape through features like real-time threat detection, predictive analysis, and automated response actions. In the video linked you can know from our experts the different modules developed: - Maialen Eceiza Olaizola from IKERLAN - Jon Egaña Zubia from S21sec - Víctor Julio Ramírez Durán, Ph.D. from IKERLAN - Saeid Sheikhi from University of Oulu - Jari Partanen from Bittium - Alexander hill from OFFIS - Institute for Information Technology 𝗖𝗼𝗺𝗽𝗿𝗲𝗵𝗲𝗻𝘀𝗶𝘃𝗲 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵 𝘁𝗼 𝗖𝘆𝗯𝗲𝗿𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆: These modules leverage cutting-edge technologies such as machine learning, distributed ledger technology, and AI-driven web crawling to deliver robust, dynamic, and automated cybersecurity solutions across various industrial sectors. 𝗠𝗲𝗲𝘁 𝘁𝗵𝗲 𝗠𝗼𝗱𝘂𝗹𝗲𝘀: AMORA: Ensures seamless integration and evaluates traceability solutions by simulating cyber-attacks, enhancing audit information, and testing data infrastructure for conformance. HEIMDAL: Focuses on real-time threat detection, monitoring communication, and system status while analyzing vulnerabilities and source code. THOR: Collects real-time data from sensors and social networks to predict and analyze threats using AI-driven techniques, providing actionable insights. ODIN: Enhances decision-making by managing security alerts, transforming them into actionable insights, and automating response actions. FRIGG: Supervises defense methods through incident simulation, defining KPIs and KRIs, and implementing dynamic visualization dashboards.  𝗔𝗯𝗼𝘂𝘁 𝗜𝗗𝗨𝗡𝗡: The IDUNN project aims to create a validated technological security framework with tools and microservices for automatic and dynamic cybersecurity operations. Our tools—AMORA, HEIMDAL, THOR, ODIN, and FRIGG—are designed to protect digital infrastructures against evolving threats. Stay tuned for more updates as we continue to enhance cybersecurity measures! For more information, visit our website https://lnkd.in/dQxWNYBq #Cybersecurity #IDUNNProject #Innovation #TechNews #AI #MachineLearning #Blockchain #CyberThreats #DigitalSecurity #IndustrialTech IKERLAN | S21sec | Fagor Arrasate S. Coop.| Cluster GAIA | Mondragon Assembly Group | OFFIS - Institute for Information Technology | CoSynth | DIN Deutsches Institut für Normung e. V. | University of Oulu | Bittium

𝗦𝗻𝗲𝗮𝗸𝘆 𝗦𝗽𝗶𝗸𝗲𝘀: 𝗨𝗻𝗰𝗼𝘃𝗲𝗿𝗶𝗻𝗴 𝗦𝘁𝗲𝗮𝗹𝘁𝗵𝘆 𝗕𝗮𝗰𝗸𝗱𝗼𝗼𝗿 𝗔𝘁𝘁𝗮𝗰𝗸𝘀 𝗶𝗻 𝗦𝗽𝗶𝗸𝗶𝗻𝗴 𝗡𝗲𝘂𝗿𝗮𝗹 𝗡𝗲𝘁𝘄𝗼𝗿𝗸𝘀 𝘄𝗶𝘁𝗵 𝗡𝗲𝘂𝗿𝗼𝗺𝗼𝗿𝗽𝗵𝗶𝗰 𝗗𝗮𝘁𝗮 We share another research on the vulnerabilities of spiking neural networks (SNNs) to backdoor attacks, particularly when processing neuromorphic data. This groundbreaking study, part of the collaborative efforts between Radboud University and Ikerlan Research Centre, delves deep into the stealthy nature of these attacks and evaluates current defense mechanisms. 𝗞𝗲𝘆 𝗜𝗻𝘀𝗶𝗴𝗵𝘁𝘀: Backdoor Triggers in Neuromorphic Data: Explored diverse triggers manipulating position and color, achieving up to 100% attack success rate with minimal impact on clean accuracy. Stealthiness of Attacks: Revealed significant stealth capabilities of potent backdoor attacks, making them hard to detect. Evaluating Defenses: Adapted state-of-the-art defenses from the image domain, uncovering their limitations and compromised performance on neuromorphic data. 𝗠𝗲𝘁𝗵𝗼𝗱𝗼𝗹𝗼𝗴𝘆: Utilized neuromorphic datasets to investigate backdoor attacks. Developed various attack strategies and assessed their impact. Adapted and evaluated defense mechanisms to enhance SNN security. 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Gorka Abad (Radboud University, The Netherlands & Ikerlan Research Centre, Spain) Oguzhan Ersoy (Radboud University, The Netherlands) Stjepan Picek (Radboud University, The Netherlands) Aitor Urbieta (IKERLAN Research Centre, Spain) https://lnkd.in/dXw3K4W4 #CyberSecurity #SNN #BackdoorAttacks #NeuromorphicData #MachineLearning #Research #RadboudUniversity #IkerlanResearchCentre #Innovation #SneakySpikes

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗧𝗦𝗧𝗘𝗠: 𝗔 𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗳𝗼𝗿 𝗖𝗼𝗹𝗹𝗲𝗰𝘁𝗶𝗻𝗴 𝗖𝘆𝗯𝗲𝗿 𝗧𝗵𝗿𝗲𝗮𝘁 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗶𝗻 𝘁𝗵𝗲 W𝗶𝗹𝗱! As part of the innovative IDUNN project, we present our latest research on #TSTEM, a cutting-edge platform designed to enhance cybersecurity through real-time collection and processing of cyber threat intelligence (CTI). 𝘈𝘣𝘰𝘶𝘵 𝘛𝘚𝘛𝘌𝘔: TSTEM (Threat Streaming and Extraction Machine) autonomously searches, extracts, and indexes Indicators of Compromise (IOCs) from various online sources. This platform is built on a containerized microservice architecture and leverages advanced technologies including: - Tweepy, Scrapy, Terraform, ELK, Kafka, and MLOps - Infrastructure as Code (IaC) for streamlined management - Custom focus crawlers for comprehensive data collection - State-of-the-art NLP models like BERT and Longformer for precise classification and entity extraction 𝘒𝘦𝘺 𝘍𝘦𝘢𝘵𝘶𝘳𝘦𝘴: - Real-Time Data Processing: Efficiently handles large volumes of data in real-time. - High Accuracy: Achieves over 98% accuracy in classification and extraction tasks within a minute. - Multi-Level Classification: Ensures precise identification of relevant information with low false positives. - Automated Infrastructure Management: Reduces human error and enhances reliability. 𝘌𝘹𝑝𝘦𝘳𝘪𝘮𝘦𝘯𝘵𝘢𝘭 𝘙𝘦𝘴𝘶𝘭𝘵𝘴: TSTEM demonstrates exceptional performance with high accuracy rates, making it a powerful tool for enhancing cybersecurity measures and protecting against large-scale cyber-attacks. 𝘈𝘶𝘵𝘩𝘰𝘳𝘴: Prasasthy Balasubramanian Sadaf Nazari Danial Khosh Kholgh Alireza B. Mahmoodi Justin Seby Panos Kostakos 𝘈𝘣𝘰𝘶𝘵 𝘐𝘋𝘜𝘕𝘕 𝘗𝘳𝘰𝘫𝘦𝘤𝘵: The IDUNN project continuously strives to advance cybersecurity through innovative solutions. By integrating advanced algorithms and machine learning models, we aim to provide robust defenses against emerging cyber threats. Stay tuned for more updates and advancements from the IDUNN project. If you have any questions or would like more information about our work, feel free to get in touch! #CyberSecurity #CTI #MachineLearning #NLP #IDUNNProject #TSTEM #AI #Innovation https://lnkd.in/d_jg7gY9

 𝗦𝗮𝗳𝗲𝗴𝘂𝗮𝗿𝗱𝗶𝗻𝗴 𝗖𝘆𝗯𝗲𝗿𝘀𝗽𝗮𝗰𝗲: 𝗘𝗻𝗵𝗮𝗻𝗰𝗶𝗻𝗴 𝗠𝗮𝗹𝗶𝗰𝗶𝗼𝘂𝘀 W𝗲𝗯𝘀𝗶𝘁𝗲 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝘄𝗶𝘁𝗵 𝗣𝗦𝗢-𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗲𝗱 𝗫𝗚𝗕𝗼𝗼𝘀𝘁 𝗮𝗻𝗱 𝗙𝗶𝗿𝗲𝗳𝗹𝘆-𝗕𝗮𝘀𝗲𝗱 𝗙𝗲𝗮𝘁𝘂𝗿𝗲 𝗦𝗲𝗹𝗲𝗰𝘁𝗶𝗼𝗻 The exponential growth of internet usage has unfortunately paved the way for the expansion of malicious activities online. Among these threats, malicious websites pose a significant risk to both individuals and corporations. To combat this, we are excited to share a new robust and efficient model for the detection of various types of malicious websites, achieving high accuracy.  𝗧𝗵𝗲 𝗣𝗿𝗼𝗽𝗼𝘀𝗲𝗱 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵: Our innovative model employs a two-step process to enhance detection accuracy: Feature Selection with Firefly Algorithm: Identifies the most relevant features, improving model efficiency and accuracy. Classification with PSO-Optimized XGBoost: Utilizes an optimized version of the XGBoost algorithm, fine-tuned using the Particle Swarm Optimization (PSO) algorithm to classify websites based on selected features. 𝗠𝗼𝗱𝗲𝗹 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻: Tested against several benchmark classification algorithms using a dataset of over 36,000 websites, our model achieved outstanding results: Binary Classification: 98.42% classification accuracy and an F1 score of 0.984. Multiclass Classification: Over 98% accuracy across each class, demonstrating robustness and reliability. 𝗞𝗲𝘆 𝗙𝗶𝗻𝗱𝗶𝗻𝗴𝘀: Our model not only demonstrates exceptional classification accuracy but also maintains high precision and minimal false error rates, making it a powerful tool for detecting various types of malicious websites and significantly enhancing cybersecurity measures. 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Saeid Sheikhi Panos Kostakos 📄 Publication Details: Type: A1 Journal article (peer-reviewed) Keywords: Cyber Security, Malicious websites, Malicious websites detection, PSO algorithm, XGBoost Published: July 3, 2024 Citation: Sheikhi, S., & Kostakos, P. (2024). Safeguarding cyberspace: Enhancing malicious website detection with PSO-optimized XGBoost and firefly-based feature selection. In Computers & Security (Vol. 142, p. 103885). Elsevier BV. DOI: 10.1016/j.cose.2024.103885 𝗔𝗯𝗼𝘂𝘁 𝗜𝗗𝗨𝗡𝗡 𝗣𝗿𝗼𝗷𝗲𝗰𝘁: The IDUNN Project continuously strives to enhance cybersecurity through innovative solutions. By integrating advanced algorithms and machine learning models, we aim to provide robust defenses against emerging cyber threats. Stay tuned for more updates and advancements from the IDUNN project. If you have any questions or would like more information about our work, feel free to get in touch! #CyberSecurity #MachineLearning #XGBoost #PSO #IDUNNProject #AI #MaliciousWebsites #Research https://lnkd.in/dxe6s8C7

𝗜𝗗𝗨𝗡𝗡 𝗣𝗥𝗢𝗝𝗘𝗖𝗧 𝗡𝗘W𝗦𝗟𝗘𝗧𝗧𝗘𝗥: 𝗜𝗡𝗧𝗥𝗢𝗗𝗨𝗖𝗜𝗡𝗚 𝗙𝗥𝗜𝗚𝗚 – 𝗧𝗛𝗘 𝗡𝗘W 𝗠𝗨𝗧𝗔𝗧𝗜𝗢𝗡 𝗦𝗧𝗘𝗣 𝗜𝗡 𝗖𝗬𝗕𝗘𝗥𝗦𝗘𝗖𝗨𝗥𝗜𝗧Y 𝗗𝗮𝘁𝗲: 22nd July 2024 In the IDUNN Project, we've developed various tools to tackle the cybersecurity challenges faced by modern industries. We're thrilled to introduce #FRIGG: The New Mutation Step in our cybersecurity framework. W𝗛𝗔𝗧 𝗜𝗦 𝗙𝗥𝗜𝗚𝗚? FRIGG is the new Mutation step in our #Cybersecurity framework cyclic process, ensuring that our defense methods produce the expected results. This involves: Defining metrics to describe the performance of deployed tools. Analyzing those metrics over time. Adjusting the tools as needed. Outlining policies for system mutation to ensure safe recovery after a cybersecurity event. W𝗢𝗥𝗞 𝗖𝗔𝗥𝗥𝗜𝗘𝗗 𝗢𝗨𝗧 𝗙𝗢𝗥 𝗙𝗥𝗜𝗚𝗚 𝘚𝘐𝘔𝘜𝘓𝘈𝘛𝘖𝘙 𝘍𝘖𝘙 𝘐𝘕𝘊𝘐𝘋𝘌𝘕𝘛𝘚 We created a comprehensive module for simulations using generative algorithm models, training machine-learning models like Hidden Markov Models, GANs, VAEs, and LDA to generate synthetic datasets for visualizing cyber-attacks. The synthetic data is securely stored in the IDA research data storage service. 𝘋𝘌𝘍𝘐𝘕𝘌 𝘛𝘏𝘌 𝘒𝘗𝘐𝘴 𝘈𝘕𝘋 𝘒𝘙𝘐𝘴 A cloud-based environment was developed for collecting and visualizing KPIs, KRIs, and KFIs. Our report, D6.4, focuses on KFIs relevant to ML models in Intrusion Detection Systems (IDS), integrating ML models with Explainable AI (XAI) frameworks. 𝘐𝘔𝘗𝘓𝘌𝘔𝘌𝘕𝘛𝘈𝘛𝘐𝘖𝘕 𝘖𝘍 𝘋𝘠𝘕𝘈𝘔𝘐𝘊 𝘝𝘐𝘚𝘜𝘈𝘓𝘐𝘡𝘈𝘛𝘐𝘖𝘕 𝘋𝘈𝘚𝘏𝘉𝘖𝘈𝘙𝘋 We developed interactive visualization widgets and automated rules through ODIN, enhancing final dashboards with explainability features via the THOR XAI interface. The FRIGG tool integrates these advancements with AMORA, HEIMDAL, and ODIN. 𝗠𝗢𝗗𝗨𝗟𝗘𝗦 𝗢𝗙 𝗙𝗥𝗜𝗚𝗚 MUTATION LOGIC ADVERSARIAL INTELLIGENCE AND MACHINE LEARNING MODELS INTERACTIVE VISUALIZATION WIDGETS IDUNN aims to validate a technological security framework composed of tools and microservices for automatic and dynamic cybersecurity operations. 𝗠𝗢𝗥𝗘 𝗔𝗕𝗢𝗨𝗧 𝗜𝗗𝗨𝗡𝗡 𝗧𝗢𝗢𝗟𝗦 The IDUNN project employs innovative tools—#AMORA, #HEIMDAL, #THOR, #ODIN, and #FRIGG—validated across three diverse industrial sectors to ensure comprehensive requirements definition and validation. Stay tuned for more updates on the IDUNN Project. If you have any questions or would like more information, feel free to reach out to us! #Cybersecurity #Innovation #IDUNNProject #FRIGG #MachineLearning #XAI #H2020 #CyberDefense #TechUpdate https://lnkd.in/dmSjue8n

 𝗘𝗻𝗵𝗮𝗻𝗰𝗶𝗻𝗴 𝗢𝗧 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘄𝗶𝘁𝗵 𝗗𝗶𝗴𝗶𝘁𝗮𝗹 𝗧𝘄𝗶𝗻𝘀 𝗮𝗻𝗱 𝗦𝗜𝗘𝗠 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻! We share a research which will help to improve security in Industrial Internet of Things (IIoT) environments. Researchers from IKERLAN Technology Research Centre and Mondragon Unibertsitatea have developed an innovative method to integrate Digital Twins (DT) with System Information and Event Management (SIEM) systems, enhancing incident response capabilities in Operational Technology (OT) environments. 𝗥e𝘀𝗲𝗮𝗿𝗰𝗵 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀: - Enhanced Threat Detection: DT-SIEM integration enables more effective real-time monitoring and threat detection. - Automated Incident Response: Leveraging Digital Twins automates and streamlines the incident response process. - Post-Incident Analysis: Facilitates comprehensive post-incident analysis and recovery, ensuring minimal downtime and continuity of operations. 𝗨𝘀𝗲 𝗖𝗮𝘀𝗲 𝗮𝗻𝗱 𝗣𝗿𝗼𝘁𝗼𝘁𝘆𝗽𝗲: A prototype and use case demonstrate the practical application and effectiveness of this integration, showcasing its potential to significantly bolster OT security against evolving threats. 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗧𝗲𝗮𝗺: Adei Arias, Cristóbal Arellano Bartolomé, Aitor Urbieta (Ikerlan Technology Research Centre, BRTA) Urko Zurutuza (Mondragon Goi Eskola Politeknikoa) Discover how this innovative approach is set to revolutionize the security landscape in IIoT environments, ensuring the resilience and continuity of industrial operations #IIoT #DigitalTwins #SIEM #Cybersecurity #OTSecurity #IncidentResponse #ResearchInnovation https://lnkd.in/dMY4r_pM

𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗛𝘂𝗻𝘁𝗚𝗣𝗧: 𝗔 𝗕𝗿𝗲𝗮𝗸𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗶𝗻 𝗔𝗻𝗼𝗺𝗮𝗹𝘆 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗘𝘅𝗽𝗹𝗮𝗶𝗻𝗮𝗯𝗹𝗲 𝗔𝗜! Article related to Idunn Project, "HuntGPT: Integrating Machine Learning-Based Anomaly Detection and Explainable AI with Large Language Models (LLMs)", co-authored by Tarek Ali and Panos Kostakos from the University of Oulu. 🔍 𝗔𝗯𝗼𝘂𝘁 𝗛𝘂𝗻𝘁𝗚𝗣𝗧: HuntGPT is a specialized intrusion detection dashboard designed to revolutionize network anomaly detection. By integrating a Random Forest classifier trained on the KDD99 dataset with powerful XAI frameworks like SHAP and Lime, HuntGPT enhances the user-friendliness and intuitiveness of anomaly detection models. 💡 𝗞𝗲𝘆 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀: Machine Learning: Utilizes a Random Forest classifier for effective anomaly detection. Explainable AI: Incorporates SHAP and Lime frameworks to provide clear and understandable model explanations. Conversational Agent: Features GPT-3.5 Turbo, delivering detected threats in an easily explainable format and offering a seamless interactive experience. 📊 𝗘𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻: We assessed HuntGPT's technical accuracy using the Certified Information Security Manager (CISM) Practice Exams and evaluated response readability across six unique metrics. Our results indicate that combining LLMs with XAI creates a robust mechanism for developing explainable and actionable AI solutions in intrusion detection systems. 👥 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Tarek Ali, Panos Kostakos 📄 𝗣𝘂𝗯𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗗𝗲𝘁𝗮𝗶𝗹𝘀: For a detailed exploration of HuntGPT’s architecture, methodology, and findings, please refer to the full article published. #Cybersecurity #MachineLearning #ExplainableAI #AnomalyDetection #IDUNNProject #LLMs #HuntGPT https://lnkd.in/dmm9XAsS