Accession Number:

AD1167570

Title:

Assured Autonomy (Briefing Charts)

Descriptive Note:

[Technical Report, Briefing Charts]

Corporate Author:

DEFENSE ADVANCED RESEARCH PROJECTS AGENCY ARLINGTON VA

Personal Author(s):

• Neema, Sandeep

Report Date:

2022-04-11

Pagination or Media Count:

25

Abstract:

Impressive progress in machine learning in last decade, but challenges remain for use in safety critical systems. Assurance and trustworthiness critical for use in safety critical systems. Program developing an assurance architecture and toolbox for assurance of learning-enabled systems, released through research community platform CPS-VO and planned through DoD platform JAIC. www.assured-autonomy.org. Established community of researchers and practitioners in AI safety. AI safety workshops, NN verification competition, CPS-VO community and design studios. Publications in top AI, CPS and Control conferences. International collaborations.

Descriptors:

• neural networks
• change detection
• computational science
• machine learning
• artificial intelligence software
• autonomous systems
• artificial intelligence
• computer languages
• reinforcement learning
• simulations
• training
• anomaly detection
• detection
• dimensionality reduction
• multiagent systems
• autonomous underwater vehicles
• autonomous vehicles

Subject Categories:

• Cybernetics
• Ground Support Systems and Facilities for Space Vehicles

Distribution Statement:

[A, Approved For Public Release]