Effective training is a vital foundation for transportation security officers required to learn strategies for identifying anomalies within X-ray images that may indicate a potential threat. Past research has shown that adaptive training is a powerful tool to increase detection performance, however, adaptive training strategies in this domain have typically utilized exposure training techniques exclusively. This paper outlines the science behind adaptive training for anomaly detection, including (1) real-time advanced performance measures associated with visual search tasks and (2) training strategies to target identified root cause(s) of error. Specific strategies discussed in this paper include exposure training and discrimination training to optimize training within the baggage screening domain. A proposed adaptive training framework and resulting system is presented.

Empirical results from a preliminary investigation into the benefits of adaptive training are presented. Thirty novice participants completed a mixed between and within design, where independent variables were training strategy (Traditional or Adaptive) and test session (Session 1, Session 2, Session 3), and dependent variables were sensitivity (d'), response criterion (c), hit rate, false alarm rate, miss rate, response time, and gaze data. In addition, eye tracking data from 4 experts was collected to evaluate differences in scan patterns and visual search strategies between novices and experts. Results showed repeated training in either group improved performance in terms of a decrease in the number of threat items missed and response time. Traditional training resulted in greater sensitivity and fewer false alarms in early training sessions. Gaze data showed that overall dwell time is positively related to the clutter density for the expert group. Analyses are ongoing to examine additional search strategy data (e.g., saccade distance, direction, changes in visual search direction, etc.) to further quantify distinct patterns in eye scan behavior to define novice versus expert performance. Future research will include further investigation into Exposure and Discrimination training to quantify benefits of each training strategy, which can better inform when and how to adapt training over time to target individualized deficiencies/inefficiencies and increase training effectiveness and efficiency. Additionally, future research should consider a longer training period, as current results did not show performance stabilization, indicating that learning may still be occurring.