Machine Learning, Reasoning and Intelligence Program

The Office of Naval Research (ONR)  Machine Learning, Reasoning and Intelligence program  is concerned with building intelligent agents that can function in the environments in which warfighters operate, that is, environments that are unstructured, open, complex and dynamically changing. Agents (cyber or physical) do not yet have the level of intelligence needed to operate in such open, uncertain and unpredictable environments either independently or alongside warfighters. The program's main objectives are to develop principles of machine intelligence, efficient computational methods, algorithms and tools for building versatile smart agents that can perform missions autonomously with minimal human supervision and collaborate seamlessly with teams of warfighters and other agents. Program focus areas include the following thrusts:

• Intelligence for Autonomous Agents: This thrust focuses on developing the intelligence needed for agents to function autonomously in a variety of situations. The following are of particular interest. (1) Building Blocks of Machine Intelligence. Some suggested topics of interest are: (a) Methods for building knowledge bases from diverse sources; (b) Learning complex concepts and tasks from examples, instructions, and demonstrations; (c) Reasoning with uncertain and qualitative information, as well as methods for meta-reasoning for self-assessment; (d) Planning in large domains in partially known environments and incompletely modeled goals and domains; (e) Intelligent architectures that seamlessly integrate knowledge-bases, learning, reasoning, and planning, for decision-making. (2) Teams of Unmanned Vehicles. Some suggested topics of interest are: (a) Computational methods for building decentralized collaborating teams of autonomous agents, in particular agents that are fairly capable in terms of sensing, communication and computational resources; (b) Mathematical theories of swarm control, particularly engineered swarms with desired behaviors. (3) Human-Agent Collaboration. Some suggested topics of interest are: (a) Multi-modal, multi-participant, human-agent dialogue systems for seamless interactions that are natural to humans; (b) Computational models of human behavior and decision-making for use by autonomous agents.

• Image Understanding: The goal of this thrust is to develop theory and algorithms for understanding surveillance imagery, for semantic search of visual datasets, and for autonomous agent perception. The main focus is on reconstructing 3D scenes, recognizing object classes and specific objects, recognizing activities and events, inferring intentions, as well as succinct natural language descriptions of images and video. Of particular interest is developing visual representations, methods for building visual knowledge bases optimized for inference, and methods for integrating reasoning with high-level knowledge and image data

Note: Proposers are encouraged to contact the program officer to discuss their research interest prior to the submission of formal proposals.

Program Contact Information

Name: Dr. Behzad Kamgar-Parsi

Title: Program Officer

Department: Code 31

Email for Questions:

* Some pages on this website provide links which require a plug in to view.