Dr. Dorin Marcu

 Associate Research Professor
 Associate Director of the Learning Agents Center (School of Computing, College of Engineering and Computing, George Mason University)

 Address: 4511 Patriot Circle, MSN 6B3, Fairfax, VA 22030
 Voice: 703-993-1535
 Email: dmarcu at gmu edu

Research


My work has contributed to research areas that are viewed as key enablers for the development of cognitive agents, and for the generalized application of artificial intelligence to complex real-world problems, including mixed-initiative human computer interaction, knowledge acquisition of problem-solving expertise from human subject matter experts, knowledge engineering, evidence-based reasoning and assumption-based reasoning. As part of my research work I have also contributed to the development of several successive generations of learning agent shells and of cognitive agents developed with them, including:

2017 – 2019
Main designer and developer of CAAPT (Cognitive Assistant for Advanced Persistent Threat Detection), a version of Cogent that can be trained by an expert cybersecurity analyst how to detect APT1 intrusions. Once trained, several copies of the agent can be configured to run in parallel in autonomous mode as trigger agent, hypothesis generation agent, automated problem-solver agent and collection service agent in an integrated intrusion detection environment.

2017 – 2018
Main designer and developer of Cogent-Arg, a version of Cogent that allows a lead analyst to elicit and integrate analytic help from a crowd of analysts.

2016 – present
Key designer and developer of sInvestigator (Science Investigator), a version of Cogent used for teaching critical thinking skills in science. The assistant helps students develop critical thinking skills in addressing scientific problems by engaging them in understanding, extending, creating, critiquing, and debating evidence-based scientific argumentations in real-life scientific investigations.

2016
Main designer and developer of CAPIP (Cognitive Agent for Persistent Intelligence Processing), a proof-of-concept cognitive agent that extends the capabilities of Cogent in the direction of automated intelligence analysis. While connected to a MITRE real-time persistent processing system in the maritime domain, CAPIP can demonstrate automated abuctive, deductive, and inductive processing in a simulated scenario.

2012 – present
Key designer and developer of Cogent, a radical simplification and improvement of both TIACRITIS and Disciple-CD.

2011 – 2014
Key developer of Disciple-CD (Connecting the Dots), an improved version of TIACRITIS with respect to usability and probabilistic theory.

2009 – 2011
Key developer of TIACRITIS, a web agent for teaching intelligence analysts the critical thinking skills needed to perform evidence-based reasoning.

1999 – present
Contributed to the development of the theory, architecture, key tools and components for the Disciple Learning Agents Shell, a suite of software tools for the development of specialized Disciple knowledge-based intelligent agents or tools, such as Disciple-WA (WorkAround plan generation), Disciple-COA (Course Of Action critiquing), Disciple-COG (strategic Center Of Gravity analysis), Disciple-LTA (Learning and Teaching Agent), Disciple-FS (Financial Services), and Disciple-EBR (Evidence-Based Reasoning).

Recognition


Books and Book Chapters


Journal Papers


Conference Papers


 

Last updated:June 30, 2020