Using behavior analysis algorithms to anticipate security threats before they impact mission critical operations

Banks, Bruce A.; Jackson, Gary M.; Helly, J.; Chin, David N.; Smith, Trevor; Schmidt, Andreas; Brewer, Paul; Medd, Roger; Masters, David; Burger, Amy; Krebs, William K.

doi:10.1109/avss.2007.4425328

Cited by 2 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The military has long been using modeling techniques for both training and predictive purposes (Banks et al, 2007), but lately even university police are using video-based predictive systems. Johns Hopkins University, for example, uses a commercially available surveillance tool called "Perceptrak" to analyze video data in real time to detect suspicious activity on its campus, such as loitering, unattended packages, and converging groups of people (Farahany, 2008).…”

mentioning

confidence: 99%

The frequency profile: An informative method for graphing the behavior of individuals post hoc or in real time.

Epstein

Mejia

Robertson

2017

Behavior Analysis: Research and Practice

View full text Add to dashboard Cite

We describe an informative type of line graph, the frequency profile, which allows a single behavior, multiple behaviors, or the specific behavioral patterns of individuals to be graphed either post hoc or in real time from digital data. The informativeness of the frequency profile curves can be optimized through the adjustment of three parameters—window size (number of adjacent time intervals in each calculation), step size (time increment by which the window is advanced), and resolution (duration of the basic observational time interval). At extreme values, the parameters yield graphical representations that have long been in use—interval recording (at one extreme) and B. F. Skinner’s cumulative record (at another extreme). Intermediate values yield informative overlapping curves that resemble probability curves, revealing (a) overall behavioral tendencies at any point in time, (b) periods of competition among different behaviors, and (c) transitions from one behavior to another. Predictive models that generate probability curves can be compared both visually and statistically with frequency profile curves. A clear graphical presentation of the moment-to-moment behavior of an individual subject should have numerous applications in business, education, law enforcement, military, and other settings in which it is advantageous to monitor individual performance continuously, in clinical settings in which subtle behavioral changes in individual clients are analyzed, or in research settings in which the behavior of individual human or animal subjects is tracked or simulated. Examples of frequency profiles with both nonhumans and humans are presented, along with statistical comparisons between a frequency profile and predicted probabilities.

show abstract

mentioning

confidence: 99%