Leveraging Symbolic Representations for Safe and Assured Learning

Abstract

Report developed under contract FA8750-19-C-0092: Leveraging Symbolic Representations for Safe and Assured Learning. This research effort targets development of novel tools, algorithms and methodologies to improve the assurance of Autonomous, Learning Enabled Cyber Physical Systems (LE-CPSs). These systems exhibit a rich set of behaviors due to higher levels of autonomy, and interaction between cyber components and the physical environment. This effort summarizes advances in symbolic system testing, model extraction, anomaly detection, learning unknown dynamics and formal approaches to verify these systems. Efforts were integrated within the Controls Systems Analysis Framework and applied to a high fidelity F16 model as the challenge problem.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 2022
Accession Number
AD1177312

Entities

People

  • Aditya Zutshi
  • Alan Fern
  • Chris Lockett
  • Suresh Jagannathan
  • Swarat Chaudhuri
  • Thomas G. Dietterich
  • Ufuk Topcu

Organizations

  • Galois, Inc.

Tags

Communities of Interest

  • Autonomy
  • Cyber
  • Energy and Power Technologies
  • Engineered Resilient Systems

DTIC Thesaurus Topics

  • Air Force
  • Airborne Collision Avoidance Systems
  • Artificial Intelligence
  • Artificial Intelligence Software
  • Automata Theory
  • Autonomous Systems
  • Bayesian Networks
  • Change Detection
  • Collision Avoidance Systems
  • Computational Science
  • Computer Languages
  • Computer Programming
  • Control Systems
  • Data Mining
  • Information Science
  • Information Systems
  • Machine Learning
  • Network Science
  • Neural Networks

Fields of Study

  • Computer science

Readers

  • Distributed Systems and Data Platform Development
  • Military Science and Technology Research and Modernization.
  • Neural Network Machine Learning.

Technology Areas

  • Cyber