Higher Automated Learning (HAL II)

The United States Marine Corps (USMC) has recognized the need for better decision making among leaders at all levels, especially among small unit leaders. In response to this the Office of Naval Research has begun a program of training research and development for Accelerated Development of Small Unit Decision Making (ADSUDM). This is a significant undertaking. In this document, Aptima proposes to develop and apply Higher Automated Learning (HAL) Phase II as a research accelerator for ADSUDM. HAL is a data mining approach that generates trainee assessments based on previous training data. It consists of a three part mathematical model to assess student progress. Principal Component Analysis identifies the knowledge components being trained and the training objectives. Hidden Markov Models assess progress along the trainee learning path. Item Response Theory produces a direct measure of trainee performance, and also identifies the difficulty of each training item. When completed, HAL will allow trainees to learn faster and retain knowledge longer, and it will allow instructors to make better feedback and remediation decisions. HAL will also help institutions reduce the costs of implementing the design and evaluation process.