(COE) Center of Neuromorphic Computing under Extreme Environments
Abstract
Incorporating Artificial Intelligence (AI) and Machine Learning (ML) techniques into Department of Defense (DoD) platforms has shown promise in enhancing information processing and operational efficiencies. However, the current reliability and robustness of existing AI-ML systems fall short of meeting the stringent demands of DoD applications in extreme environments. These environments, characterized by factors such as high temperatures, radiation exposure, corrosion, and erosion, pose unique challenges for both hardware platforms and the learning algorithms of intelligent systems. The physical platform ( hardware ) must withstand these extreme conditions, necessitating innovations in materials and device design. Simultaneously, the algorithms ( software ) must demonstrate adaptability to uncertain and changing harsh environments, aligning with the principles of neuromorphic systems. Neuromorphic computing seeks to replicate the information-processing capabilities of the human brain. However, present artificial neural networks lag significantly behind their biological counterparts in effectiveness and reliability, and their task limitations are evident when compared to the human brain s versatility. This disparity is at least partially attributed to the failure of current artificial neural networks to capture key dynamics of the brain, many of which are derived from the movement of ions across biological membranes and synapses. Replicating such dynamics in inorganic systems, more resilient to extreme environments than biological tissues, offers a promising avenue to achieve reliable and efficient intelligent systems tailored to future DoD requirements.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 06, 2025
- Source ID
- FA95502410322
Entities
People
- Joshua Yang
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Southern California