MD and FL-Human Research & Engineering Dir-FFP, HRED/Simulation and Training Technology Center - FFP
|Sottilare, Robert Anthony
The Army Research Laboratory’s adaptive computer-based tutoring research program is conducted in Orlando as part of the Learning in Intelligent Tutoring Environments (LITE) Laboratory program. This goal of this research is to develop methods to assess the cognitive and affective states of trainees in near-real-time and then use this state data to adapt computer-based instruction to enhance/accelerate learning outcomes (e.g., knowledge and skill acquisition, retention). The research program has three thrusts: trainee modeling, authoring and expert modeling, and instructional strategy selection. Trainee modeling includes assessment techniques for determining cognitive and affective states through behavioral and physiological sensing techniques and machine learning algorithms. Authoring and expert modeling research explores the automated development of instructional content and expert models used as standards to define the trainee’s performance level. Finally, instructional strategy selection research develops and assesses machine learning techniques to automatically guide the tutor’s performance (e.g., interaction with the trainee, selection of instructional content and feedback, pace, and challenge level of instruction).
Researchers will have the opportunity to work closely with adaptive tutoring scientists in developing, applying, and assessing new automated tutoring technologies (tools and methods). Emphasis in this program is on creating open-source architecture to facilitate the authoring, use, and assessment of adaptive tutoring methodologies through literature reviews, market reviews, and experimentation. This server-based architecture is the Generalized Intelligent Framework for Tutoring (GIFT - http://www.GIFTtutoring.org/). Experiments include human subjects locally and at locations that include, but are not limited to the United States Military Academy (USMA) and the Infantry School at Ft. Benning, Georgia. Significant interaction with academia (e.g., USC Institute for Creative Technologies, the UCF Institute for Simulation and Training, and the University of Memphis) can be anticipated through cooperative agreements and public conferences.
For the next evolution of GIFT, we are interested in developing Markov Decision Logic Networks to support modeling of both learner and tutor actions across a variety of training environment states.
Sottilare R: Proceedings of the 10th International Conference on Intelligent Tutoring Systems 6095: 411, 2010
Sottilare R: Considerations in the development of an ontology for a Generalized Intelligent Framework for Tutoring. International Defense & Homeland Security Simulation Workshop in Proceedings of the I3M Conference. Vienna, Austria, September 2012
Intelligent tutoring systems; Markov decision processes; Instructional strategy selection; Adaptive tutoring; Trainee modeling; Machine learning; Authoring and expert modeling;