Foundational Machine Intelligence*
Abstract
*Previously in Foundational Computer Science. (U) The Foundational Machine Intelligence program is supporting research on the foundations of artificial intelligence and machine learning and reasoning. One focus is on techniques that can efficiently process and “understand” massive data streams. Deeply layered machine learning engines will be created that use a single set of methods in multiple layers (at least three internally) to generate progressively more sophisticated representations of patterns, invariants, and correlations from data inputs. These will have far-reaching military implications with potential applications such as anomaly detection, object recognition, language understanding, information retrieval, pattern recognition, robotic task learning and automatic metadata extraction from video streams, sensor data, and multi-media objects. Foundational Machine Intelligence also examines the human aspects of computing, with interest in collaboration, interaction and information exchange; non-symbolic representation/reasoning paradigms based upon a universal “cortical” algorithm; unmanned vehicles and intelligent agents that generate and manage their own goals within human-described mission constraints; and modeling of human language acquisition by associating words with the real-world entities perceived through multiple modes of sensory input.
Document Details
- Document Type
- Accomplishment
- Publication Date
- Oct 01, 2011
- Source ID
- 669343f2c3321a74fd8d1835b70da7d4