Cognitive Computing

Predefined commands for computers tend to overlook unforeseen circumstances that would render such commands ineffective. Cognitive computing solves this problem by programming military systems to operate autonomously using learning and reasoning to function in a sustainable manner.

Objectives

Future military systems must operate and interact at the semantic level if they are to perform increasingly complex and integrated functions in the network-centric battlespace. Ontology-driven interpretations and cognitive processing are critical enablers of semantic operation and integration.

Predefined computation is insufficient for fully autonomous systems, where unforeseeable and potentially catastrophic events may occur. Complex, cognitive processing for real-time, autonomous systems requires run-time, adaptive discovery, composition, and execution of appropriate behaviors in a computably sustainable manner.

Key Technologies

We are developing prototype applications for ontology-driven interpretations and cognitive processing. Our key focus areas include:

• Polymorphous cognitive agent architecture to:
  • Reduce complexity using a cognitive layer to dynamically and adaptively map applications to cognitive components.
  • Improve scalability through a reconfigurable agent virtual machine layer that can adaptively and efficiently reconfigure massively parallel processors.
  • Enable efficient processing of heretofore prohibitively complex cognitive computation.
• Ontology translation protocol (Ontrapo) provides:
  • Automated processes for aligning meta-data descriptions of heterogeneous information resources.
  • Alignment algorithms to dynamically and automatically discover semantic correspondences between the elements of ontologies and schemas.
  • Leadership in the research community by organizing ontology conferences and workshops.
• Analogical experience sharing ontological protocol (AESOP) for:
  • Techniques for transferring expertise between agents to accelerate learning in new environments.
  • Distributed multi-agent configuration optimmziation in multiple domains.

Applications and Transition Opportunities

Cognitive-computing technology will lead to a new generation of computing hardware and software that will enable revolutionary advances in cognitive-information processing for real-time defense and intelligence applications. This area will be a significant technology for us in the next three to seven years. Potential technology transfer-targets include:

• Unmanned Combat Armed Rotorcraft
• Joint Unmanned Combat Air Systems
• Intelligence Analysis.

Cognitive Computing is one of several Business Areas for the Distributed Processing Laboratory