© 2016 krig research

Synthetic Visual Cortex Model for Visual Learning

The NeuroMatrix(tm) system contains a complete model of the visual cortex of the human brain, including models of the eyes, early visual processing centers, and higher level reasoning centers. NeuroMatrix(tm) advances beyond deep learning into volume learning, capable of learning precise signatures of individual objects, rather than generic object signatures via deep learning.


NeuroMatrix(tm) provides three major components:


  • Visual Neurocortex Model: A common model of the human visual system from the eyes through the early visual processing centers, using volume learning, accessed by the higher level learning agents.
  • Autonomous Learning Agents:  Individual agents can simultaneously explore and classify visual environments using self-directed learning. Each agent is independent, and can be directed to discover specific types of visual items. Agent discoveries can later be reviewed with experts for final naming and classification. Agents store their knowledge in shareable downloads.
  • Matrix Downloads: Similar to the intelligence downloads in the movie ''The Matrix', the intelligence learned by agents is stored, and can be later recalled and downloaded into another NeuroMatrix(tm) system to augment intelligence for specific tasks.


NeuroMatrix(tm) can be deployed to learn in controlled and uncontrolled environments. Agent learnings can be exchanged via downloads between NeuroMatrix(tm) systems. Agent downloads can be purchased from Krig Research for specific visual domains. The NeuroMatrix(tm) system operates on a client computing device, supported by a server device. The server device may be remote or local.



Evades Electronic Warfare and Jamming countermeasures


The NeuroTarget(tm) system operates on visible frequency light

images, therefore is capable of surviving common electronic warfare countermeasures which affect radar-based systems.


NeuroTarget(tm) provides two major components:


  • Target Signature Analysis: Learns a target signature, and then provides constant updates on material changes to the target, such as weapon strike and explosive damage. For example, a target for small weapon firing accuracy analysis can be learned, or a target may be a larger object such as a tank, airplane, or building.
  • Weapons Firing Events: Analyzes weapons firing events, such as small arms round firing or larger projectile firing, to locate actors who have shot the target, including shooter location, range, and projectile trajectory details.


NeuroTarget(tm) can be customized to analyze a range of targets, and a range of weapons firing events.


NeuroMatrix(tm) downloads provide domain intelligence

Neurotarget (tm)

neuromatrix (tm)

Synthetic neuron clusters contain visual genomes recorded by agents during volume learning

world class expertise              subcontracting              consulting

Products

SYNTHETIC VISION           COMPUTER VISION           IMAGING