Data Analytics for Cyber Security: Defeating the Active Adversaries

Abstract

The overall arching goal of this project is to develop techniques to increase the robustness of machine learning systems against adversarial attacks. In addition, we focus on understanding how the developed techniques could be leveraged in practice so that humans and ML tools could work together. Accomplishments: Please see the attached report and slides. Training Opportunities: We have organized a reading group on deep learning and adversarial machine learning topics allowing students to learn the recent advances in this area. In addition, Dr. Kantarcioglu thought a course on adversarial machine learning. The students who participated in this research had chance to learn cutting edge technologies with respect to deep learning, adversarial machine learning, IoT data, and explainable AI. Results Dissemination: The results are disseminated via publishing our work in major conferences and venues. In addition, PI Kantarcioglu and co-PIs Thurasingham and Xi gave many talks on the topic. Co-PI Thuraisingham participated in Women in Data Science event held Stanford University that was live streamed to around 100,000 people. She gave an overview of our Adversarial Machine Learning Research for ARO project.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 06, 2023
Accession Number
AD1222215

Entities

People

  • Murat Kantarcıoğlu

Organizations

  • University of Texas at Dallas

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Brain and Cognitive Science; Experimental Psychology; Cognitive Neuroscience
  • Distributed Systems and Data Platform Development

Technology Areas

  • 5G
  • 5G - DoD 5G Program
  • 5G - Internet of Things
  • AI & ML
  • AI & ML - Neural Networks
  • Cyber