EMERging concepts and innovative technologies

Does Game-Based Learning Work? Results from Three Recent Studies

Fact or Fiction - Soldiers are Gamers:  Potential Effects on Training

The Utilization of Low Cost Gaming Hardware in Conventional Simulation

Eye Limited Resolution Displays: On The Cusp?

Task Validation of Display Temporal-Resolution Measurements

Automated Shader Generation using a Shader Infrastructure

Building Interactive Virtual Humans for Training Environments

Intelligent  Agents:  Incorporating  Personality  into  Crowd  Simulation

Training Approaches for Honing Junior Leader Adaptive Thinking,  Cultural Awareness and Metacognitive Agility

ASSESSMENT OF TWO DESK-TOP COMPUTER SIMULATIONS  USED TO TRAIN TACTICAL DECISION MAKING OF  SMALL UNIT INFANTRY LEADERS

High Fidelity Flight Training Devices for Training Ab Initio Pilots

DO BETTER MULTI-TASKERS MAKE BETTER PILOTS?

Use of Haptic Devices to Provide Contextual Cues in a Virtual Environment for Training

Markerless Augmented Reality based Cameras using System-on-Chip Technology

Synergistic Classifier Fusion for Security Applications

Learner Assessment Data Models for standardizing assessment across Live, Virtual and Constructive Domains

Mixed Initiative Team Performance Assessment System (MITPAS) For Training and Operation

 

 

 

Does Game-Based Learning Work? Results from Three Recent Studies

2007 Paper No. 7172

 

Richard Blunt, Ph.D.

DeVry University

Arlington, VA

 

The Department of Defense (DoD) is faced with challenges in expanding technology-based solutions that can make Warfighters more efficient, effective, knowledgeable, and flexible.  Of growing important to the DoD is the potential of using Commercial Off the Shelf (COTS) game-based learning in the armed forces for increasing combat readiness.  The recruits of today not only understand technology in everyday use, they expect it.  These young recruits are “digital natives” who were raised in a digital environment surrounded by inexpensive, yet highly interactive gaming systems.  To get the most from our new “best and brightest,” new research into game-based learning needs to focus on military use.  The objective of these projects was to add definitive research in the badly needed area of game-based learning.    Three research studies were conducted at a national university to examine the difference in academic achievement between students who did and did not use video games in learning. Three different video games were added to approximately half the classes of freshmen Introduction to Business and Technology courses, 3rd year Economics courses, and 3rd year Management courses. Identical testing situations were used in all courses while data collected included game use, test scores, gender, ethnicity, and age. ANOVA, chi-squared, and t tests were used to test game use effectiveness.  Students in classes using the game scored significantly higher means than classes that did not. There were no significant differences between genders, yet both genders scored significantly higher with game play. There were no significant differences between ethnicities, yet all ethnic groups scored significantly higher with game play. Students 40 years and under scored significantly higher with game play, while students 41 and older did not.  These studies add definitive research in the badly needed area of game-based learning.  The DoD now has studies proving the efficacy of digital game-based learning and how it can improve learning.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

 

Fact or Fiction - Soldiers are Gamers:  Potential Effects on Training

2007 Paper No. 7341

 

James Belanich, Daniel B. Horn, Jennifer L. Solberg

U.S. Army Research Institute for the Behavioral and Social Sciences

Arlington, VA

 

Karin A. Orvis

Old Dominion University & Consortium Research Fellows Program

Norfolk, VA

 

Jennifer C. Moore

George Mason University & Consortium Research Fellows Program

Fairfax, VA

 

PC-based games are increasingly being used for military training domains. Proponents of training games argue that younger Soldiers are part of the “digital” generation, and having grown up playing videogames they will respond positively to their use in training. However, in a series of research projects we’ve found that these assumptions may be overstated. This paper covers two research efforts that analyzed Soldiers’ videogame experience, as well as the impact of trainees’ prior experience on training outcomes. The first project surveyed a total of 777 first-year U. S. Military Academy Cadets who participated in a team tactics training exercise using America’s Army, over two years. Results across both years indicated that 60% of Cadets had limited or no videogame experience in the prior year. Additionally, the amount and type of prior gaming experience correlated with training outcomes (i.e., performance, training satisfaction, motivation, and time on task). A second project assessed the frequency that Soldiers of all ranks play videogames compared to engagement in other recreational activities.  Results suggest that fewer than 32% of over 10,000 U.S. Army Soldiers surveyed across various ranks play videogames recreationally on a weekly basis (numbers vary by rank). For the ranks with the highest frequency (E2-E4), only…

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

 

The Utilization of Low Cost Gaming Hardware in Conventional Simulation

2007 Paper No. 7256

 

Peter Smith, Lee Sciarini, Denise Nicholson

Institute for Simulation and Training Orlando, Fl

 

The current convergence between conventional simulation training and game based training is blurring the lines that once defined each industry. The simulation industry has already begun to assimilate low cost gaming middleware for graphics, artificial intelligence, and physics. With the unequivocal success of the integration of game based software solutions into conventional training , the next logical step is the utilization of low cost gaming console hardware within the tool chain.   The most recent generation of gaming console hardware is ripe with potential uses within simulation and training. The three major game console manufacturers, Microsoft, Nintendo, and Sony, have each provided open tools that allow direct access to their various hardware solutions. The potential of these systems envelop a broad range of uses from deploying fully immersive 3D simulations and performing computationally intensive multithreaded tasks to revolutionary paradigms for user interaction.    Determining the appropriate solution for a given situation can be a daunting task for uninitiated developers, as each solution is unique to its hardware. Given this, training objectives and requirements must be mapped to the full range of fidelity and capabilities offered by each solution. A recent study to determine the utility of low cost gaming hardware in the Deployable Virtual Training Environment (DVTE) was conducted by a multidisciplinary team from the Applied Cognition and Training in Immersive Virtual Environments (ACTIVE) Lab at the University of Central Florida’s Institute for Simulation and Training. This work discusses the initial findings from the study and provides guidelines for employing the latest generation of gaming hardware within PC Simulation and Training. 

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Eye Limited Resolution Displays: On The Cusp?

2007 Paper No. 7075

 

Harry Streid

Boeing Training Systems and Services

St Louis, MO

 

As the cost per pixel of image generation and display hardware has fallen, it has become economically feasible to consider constructing a wide field of view visual system having eye-limited resolution.  Many simulation and training visual systems have been fielded over the years with resolution barely equivalent to legally blind levels of acuity, so it is certainly a welcome opportunity to improve training capability by bringing these new technologies to bear.  It is hard to argue that incrementally better resolution is not always incrementally better, but trade-offs made simply to provide more pixels to the display  may actually reduce overall image quality.  This paper focuses on identifying those trade-offs and how they must be managed so that truly eye-limited displays can be fielded.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Task Validation of Display Temporal-Resolution Measurements

2007 Paper No. 7274

 

Marc D. Winterbottom

Air Force Research Laboratory

Meza, AZ

 

George A. Geri, Lisa Park

Link Simulation and Training

Mesa, AZ

 

We have recently described a perceptual technique for assessing the temporal resolution of display systems used in Air Force flight-simulators. That technique was based on the assessment of the perceived blur of a simple stimulus consisting of a pair of moving vertical lines. In the present study, we have attempted to validate our previous technique by correlating it directly with performance on simplified tasks that may be performed during training in a high-fidelity Air Force flight simulator. Data were obtained by asking observers to; 1) detect whether or not a moving F-16 target-aircraft banked as it moved laterally across the observers’ field-of-view and 2) judge whether changes in aircraft pitch resulted in blurring of the associated moving terrain. The level of moving-image blur was determined by the length of time that the image was presented during each video frame (i.e., the project or hold-time) on a digital (DLP) projector. The results were compared to those obtained using a standard CRT projector, whose effective hold-time was about one-quarter of that of the lowest DLP hold-time tested. For both the roll detection task and the aircraft pitch task, results with the DLP project or did not differ significantly from that obtained with the CRT when the DLP hold-time was reduced to 5.8 msec. These results are in qualitative agreement with those obtained using the simpler moving-line test, and suggest that the latter is a valid measure of display temporal resolution in the context of flight-simulator applications.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Automated Shader Generation using a Shader Infrastructure

2007 Paper No. 7097

 

Chris Coleman, Kevin Harris, Anthony Hinton, Anton Ephanov

MultiGen-Paradigm

Richardson, TX

 

Programmable rendering on the GPU (Graphics Processing Unit) utilizing “shader” technology has become a recognized benefit to visual simulation, providing unprecedented realism and fidelity to synthetic environments. However, effective use of shaders is a technological challenge, where implementation of even a trivial vertex or fragment shader requires manual re-implementation of the fixed-function pipeline. Furthermore, generation of efficient shaders for existing data is a complicated undertaking that is compounded by the number of state permutations possible under a modern graphics API, such as OpenGL, and the desire to extend fixed function state with advanced rendering techniques.  Therefore, system integrators and run-time providers face significant challenges in incorporating shader technology while hiding complexity and maintaining backwards compatibility with existing data. Preserving the massive investment in existing database and model libraries, while enhancing system capabilities, is a fundamental concern. Though some shader techniques are required for modeling, others are typically supplied by the run-time. Fundamentally, it is clearly disadvantageous to have run-time code contained in shared modeling assets.  As a solution to these problems, this paper introduces a Shader Infrastructure that automates the building of vertex and fragment shaders by analyzing the OpenGL state machine. The Shader Infrastructure is capable of not only dynamic generation of highly efficient shaders for rendering any legacy data considered valid by the OpenFlightTM standard, but also extending the rendering pipeline via Advanced Rendering Techniques. Techniques allow for implementation and merging of novel GPU-based rendering approaches by injecting small snippets of shader code into the Shader Infrastructure. The Shader Infrastructure allows designers to customize portions of the rendering pipeline and to automatically combine that customization with other rendering techniques, either fixed-function or shader-based. Therefore, it significantly simplifies the problem of content management and reuse while taking full advantage of the advances in the programmable PC…

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Building Interactive Virtual Humans for Training Environments

2007 Paper No. 7105

 

Patrick Kenny, Arno Hartholt, Jonathan Gratch, William Swartout, David Traum, Stacy Marsella, Diane Piepol

Institute for Creative Technologies / University of Southern California

Marina Del Rey, CA

 

There is a great need in the Joint Forces to have human to human interpersonal training for skills such as negotiation, leadership, interviewing and cultural training. Virtual environments can be incredible training tools if used properly and used for the correct training application. Virtual environments have already been very successful in training Warfighters how to operate vehicles and weapons systems. At the Institute for Creative Technologies (ICT) we have been exploring a new question: can virtual environments be used to train Warfighters in interpersonal skills such as negotiation, tactical questioning and leadership that are so critical for success in the contemporary operating environment?  Using embodied conversational agents to create this type of training system has been one of the goals of the Virtual Humans project at the institute. ICT has a great deal of experience building complex, integrated and immersive training systems that address the human factor needs for training experiences.  This paper will address the research, technology and value of developing virtual humans for training environments. This research includes speech recognition, natural language understanding & generation, dialogue management, cognitive agents, emotion modeling, question response managers, speech generation and non-verbal behavior. Also addressed will be the diverse set of training environments we have developed for the system, from single computer laptops to multi-computer immersive displays to real and virtual integrated environments.   This paper will also discuss the problems, issues and solutions we encountered while building these systems. The paper will recount subject testing we have performed in these environments and results we have obtained from users. Finally the future of this type of Virtual Humans technology and training applications will be discussed.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

 

Intelligent Agents: Incorporating Personality into Crowd Simulation

2007 Paper No. 7151

 

Sivakumar Jaganathan, Dr. T. L. Clarke, Dr. D. J. Kaup, Jitendra Koshti, Dr. L.C. Malone, Rex Oleson

University of Central Florida

Orlando, FL

 

The simulation and modeling of crowd behavior has become an active research area in recent years. This area of research has been applied to a wide variety of domains such as military, education, training, entertainment and human factors analysis. Our current research at the Institute for Simulation and Training (IST) which is a part of University of Central Florida (UCF) focuses on the influence of individual personality factors on crowd behavior. The objective of this research is to model and simulate the “Big Five” personality factors and associated forces using the Helbing-Molnar-Farjas-Vicsek (HMFV) crowd model within the MASON (Multi-Agent Simulation of Neighborhoods) environment. MASON is a fast discrete-event multi-agent simulation library with the core of the programming written in Java. It provides wealth of functionality for many lightweight simulation needs and the ability to record simulation results for comparison with real data.  The crowd behavior models are validated by comparison with real-life video recorded at venues such as churches and sporting events.  This comparison is facilitated by use of in-house software developed using the OpenCV (Open Computer Vision) library to derive optical flow from the videos. The HMFV model is being enhanced by the incorporation of personality factors, and the coefficients associated with the personality factors are being optimized by comparison against the real-life video using the in house developed software tool.  The results of the comparison are discussed.  This research falls within the technical area of modeling and simulation of the behavior of individuals in teams and in crowds and other social groups. It will find application in a variety of areas such as emergency management of crowds and for the design of buildings for emergency egress. An additional area of application of such models of collective human behavior is to identify anomalous crowd behavior associated with the Improvised Explosive Device (IED) threat.  

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Training Approaches for Honing Junior Leader Adaptive Thinking,  Cultural Awareness and Metacognitive Agility

2007 Paper No. 7467

 

Elaine M. Raybourn, Ph.D.

Sandia National Laboratories

Albuquerque, NM

 

The present paper is the third in a series that seeks to describe the theories and methods employed to create engaging learning environments for training Marine and Army junior leaders, U.S. Special Forces, Civil Affairs, and Psychological Operations teams to think adaptively. The first publication (I/ITSEC 2005) described a novel use of a first-person shooter game-based training system (in use since 2004) at the JKF Special Warfare Center and School that focuses on negotiation skills, cross-cultural communication. The second publication (I/ITSEC 2006) described the simulation experience design method used to create a “crucible” experiences that invoke trainee adaptive thinking by forcing crucial choices, and sharpening one’s focus. The present paper addresses more specifically the perspectives that have inspired the development of methods (used first for the U.S. Special Forces and currently for DARPA DARWARS Ambush NK! [Non-Kinetic]) to train adaptive thinking particularly by honing cultural awareness and metacognitive agility for non-kinetic engagements.  In the full paper, we discuss perspectives from intercultural communication, social-process simulation, and metacogntion that have inspired the approach to overall training architecture and software development to train adaptive thinking, cultural awareness, and metacognitive agility for multi-player game-based systems. We describe how game-based training can be designed as consisting of a system of experiences, and how the design of a reflective role, in-game assessments & evaluation, and quantitative evaluations in after action reviews enhanced for non-kinetic engagements, present a unique blend of methods from which to enhance adaptive thinking. The paper addresses how the instantiated role functionality and methods can be used by observer controllers, peer trainees, subject matter or cultural experts, instructors, etc. to provide quantitative feedback of actions taken, (including communications) as they occur in real-time. We discuss how our approach instantiates in software a unique role that provides experience with…

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

ASSESSMENT OF TWO DESK-TOP COMPUTER SIMULATIONS  USED TO TRAIN TACTICAL DECISION MAKING OF  SMALL UNIT INFANTRY LEADERS

2007 Paper No. 7007

 

Dr. Scott A. Beal

U.S. Army Research Institute

Fort Benning, GA

 

Fifty-two leaders in the Basic Non-Commissioned Officer Course (BNCOC) at Fort Benning, Georgia, participated in an assessment of two desk-top computer simulations used to train tactical decision making. Thirteen leaders trained with the Soldier Visualization Station (SVS) simulation, while 39 others trained with Simulation Field Exercise, or SimFX. Pre-simulation exercise measures included military and simulation experience, and tactical situation judgment. A questionnaire administered to leaders following simulation exercises documented their sense of personal involvement during mission execution and their perceptions of the training value of the simulations.  Leaders in both groups were assessed individually for their ability to recognize and implement rapid, sound tactical decisions while serving as a leader of a light Infantry squad during urban patrol and defense missions. Results suggested that desk-top simulations can be used to train the tactical decisions leaders make during exercises that require greater expenditures of resources. The methods used to train with simulations impacted leaders’ tactical decision making, their perceptions of the training value of the simulations, and their ideas about what they learned from the experience. Instructor-facilitated training resulted in better tactical decision making. 

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

High Fidelity Flight Training Devices for Training Ab Initio Pilots

2007 Paper No. 7145

 

Nickolas D. Macchiarella, Shawn M. Doherty

Embry-Riddle Aeronautical University

Daytona Beach, Florida

 

An experimental flight training program at Embry-Riddle Aeronautical University (ERAU) investigated the degree to which flight simulation could be used in a training curricula for ab initio pilots.  The experimental group earned an FAA private pilot’s certificate using a curriculum largely comprised of simulated flight in flight training devices (FTD); the control group trained exclusively in airplanes.  What sets this research apart from prior transfer of training (ToT) work is the high level of simulator usage, (60% FTD-based flight) and the FAA’s approval for certifying pilots with this methodology. Researchers hypothesized that optimizing the application of specifically designed FTDs will afford the simulation related training benefits typically associated with more costly “graduate level” flight training (e.g., the FAA’s Advanced Qualification Program [AQP] and military advanced aircraft type ratings) to ab initio flight training for a relatively low cost.  At the core of this research investigation was a classic transfer of training (ToT) study examining 34 flight tasks.  Transfer can be measured using the transfer effectiveness ratio (TER) equation.  The results from the amount of transfer from the FTD to actual aircraft flight suggested implications for both adjustments to the flight training curricula and for specific modifications to the FTD as applied in an ab-initio training program. More specifically, these results provided an indication that added visual fidelity, in terms of graphical 3D artwork, was necessary in the virtual environment for particular ground reference maneuver tasks.  A low fidelity visual scene at low level flight altitudes provided poor cues for pilots training for ground reference maneuvers.  Additionally, the level of traffic in the scenario and degree of complexity in simulated airspace affected transfer to the real world flights.  

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

DO BETTER MULTI-TASKERS MAKE BETTER PILOTS?

2007 Paper No. 7193

 

V. Alan Spiker, Tricia Mautone, Susan Fischer

Anacapa Sciences, Inc.

Santa Barbara, CA

M. Ron Karp

Arizona State University

Mesa, AZ

 

Aircraft operation involves many facets of multi-tasking (MT), where breakdowns in task management have serious implications for performance and safety. There is a need to develop valid, predictive tests of MT ability and provide practical criterion measures of that performance. Thirty advanced students (15 pilot-copilot pairs) in a university flight training program were tested in a medium-fidelity King Air simulator. The 30-min. scenario was designed to task-load both pilot and copilot during an event-filled instrument approach to an unpublished holding point, challenging procedure turn, and steep descent with loss of glideslope and worsening weather, culminating in a missed approach. Two observers independently scored subjects’ performance on 65 MT-relevant behavioral events and rated them on six process dimensions of MT. Subjects also took a battery of predictive tests, including two specifically-designed MT tests, and tests of fluid intelligence, processing speed, and aviation mathematics. The MT tests assessed ability to simultaneously monitor multiple visual fields and hold multi-dimensional concepts in memory. The scenario challenged MT for all crews, with 40% of the events showing evidence of disruption. Inter-observer reliability for MT ratings and sub-event performance was high (r=.85-.91) and MT test reliability coefficients exceeded .80. MT criterion measures showed a complex, but fascinating, relationship to the predictive tests. One test of MT was significantly related to performance for copilots but not pilots. The other test of MT predicted performance on two key archival measures of student proficiency (hours to instrument rating, number of extra flights). The paper closes with study implications for developing other criterion measures of MT, adapting MT tests for student placement, and developing MT training programs for “at-risk” student-pilots.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Use of Haptic Devices to Provide Contextual Cues in a Virtual Environment for Training

2007 Paper No. 7227

 

Amanda Hafich, Jennifer Fowlkes, Patrick Lenihan

CHI Systems, Inc.

Orlando, FL

 

The usefulness of incorporating haptic devices into virtual environments (VEs) is becoming an increasingly prevalent area of research in the training arena.  The work described in this paper investigates the utility of incorporating haptics in a VE for Military Operations in Urban Terrain (MOUT) training, focusing on two primary research questions:  How do haptic cues interact with and compare to auditory cues for the identification of task-relevant VE events, and how does the inclusion of haptic cues impact presence using the Witmer & Singer (1998) Presence Questionnaire? To answer these questions, participants performed a building search task eight successive times in a desktop VE.  Using vibrotactile stimulation applied to the torso, arms and legs, participants received either no haptic cues, haptic cues that were approximations of real-world stimuli (“Natural Haptics”), or haptic cues which were metaphoric signals for real-world stimuli (“Metaphoric Haptics”).  Across the haptics groups, half of the participants were given auditory cues for the stimuli events (explosions, gunshots, and collisions), and the other half received no auditory stimuli.  All participants received identical visual cues.  Results from 74 participants showed that participants in the Natural Haptics condition showed better identification of collision and gunshot events compared to participants in the other groups, whereas participants receiving auditory cues showed better identification of explosion events.  Regarding presence, the scores for participants in the Natural Haptics group only increased over trials on the Interface Quality subscale, and participants in both haptic groups showed increased scores over trials for the Adaptation and Immersion subscale. However, there was not a general advantage in presence scores that was related to haptics, perhaps due to ceiling effects. Other findings showed that haptics was associated with increased perceived task demands.  

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Markerless Augmented Reality based Cameras using System-on-Chip Technology

2007 Paper No. 7257

 

Veronica Teichrieb, João Marcelo X. N. Teixeira, João Paulo S. do M. Lima, Judith Kelner

Federal University of Pernambuco – UFPE, Computer Science Center – CIn

Recife, Pernambuco, Brazil

 

Markerless augmented reality (MAR) aims to integrate 3D virtual objects into the real environment in real-time, enhancing the user’s perception and interaction. It differs from marker based augmented reality (AR) systems basically by the method used to place virtual objects in the real world. In MAR, the real environment may be used as a marker that can be tracked in order to position virtual objects. Generally, the techniques used are based on computer vision, image processing, and computer graphics, and a major issue related to the field is robust, yet precise real-time object tracking and registration. Several AR solutions using general purpose devices have been developed. Such processing is done by software, making it difficult to obtain real-time results without compromising resolution and frame rate, and requiring the use of high clock frequencies that consequently lead to higher costs and power consumption. This paper introduces MARCam, an FPGA based solution that allows the development of embedded MAR applications. This framework allows the development of compact hardware with wearable capabilities for applications requiring user mobility through unknown environments and real-time dedicated processing. One of the most promising applications of this technology is guidance in training systems. MARCam can utilize a Structure from Motion (SfM) based technique. Instead of relying on previously obtained information about the real scene, SfM based techniques estimate the camera displacement without a priori knowledge of the environment. These methods are also able to retrieve the structure of the scene in real-time, with various levels of detail. Due to this, it is possible to reconstruct a totally unknown environment on the fly. MARCam’s architecture is divided into many circuit modules, each one responsible for a specific task. This way, one can quickly arrange an assembly of modules and have in short time a fully-working, dedicated MAR system.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Synergistic Classifier Fusion for Security Applications

2007 Paper No. 7489

 

Paul F. Evangelista

United States Military Academy

West Point, NY

Mark J. Embrechts, Boleslaw K. Szymanski

Rensselaer Polytechnic Institute

Troy, NY

 

Unbalanced, high-dimensional, binary classifications create challenges in a variety of systems and environments, most notably within physical security and computer security domains.  The imbalance within these problems consists of a significant majority of the negative (healthy, non-intruding) class and a minority (unhealthy, intruding) positive class.  Any system that needs protection from malicious activity, intruders, theft, or other types of breaches in security must address this type of problem.  Given numerical data that represent observations or instances which require classification, many practitioners apply state of the art machine learning algorithms to aid in solving unbalanced classification problems.  The unbalanced and high-dimensional structure of the data can trouble these learning methods. High-dimensional data poses a ``curse of dimensionality'' which can be overcome through subspace modeling and intelligent fusion.  A fundamental method for evaluation of the binary classification model is the receiver operating characteristic (ROC) curve and the area under the curve (AUC), and the intelligent fusion employed ties directly with the properties of this evaluation method.  This work exposes the underlying statistics involved with ROC curves and leverages these properties to create synergistic classifier fusion through rankings.  Decision ROC charts are a novel illustration that augment the ROC curve to provide a more complete representation of the classifier performance.  Pseudo-ROC curves, created from simulated rankings utilizing principles based on the Wilcoxon-Rank sum or Mann-Whitney U statistic, provide novel insight into the behavior of classifier rankings.  The critical finding involves the unique behavior of rankings for unbalanced classification problems and methods to capitalize on this behavior to improve classifier accuracy for unbalanced problems.  Arguments presented include theoretical discussion, proof of principle through simulated classifier rankings examined with a factorial design, and experimental results with actual data including host-based and network-based intrusion detection datasets.

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

Learner Assessment Data Models for standardizing assessment across Live, Virtual and Constructive Domains

2007 Paper No. 7366

 

Brent Smith, Chris DuBuc

ECS

Orlando, FL

Dean Marvin

Joint ADL Co­Lab

Orlando, FL

 

Feedback and guidance is an essential component of any learning environment so that errors in performance can be pointed out and corrected. A crucial part of this process is the ability to identify gaps in skills or knowledge and assess the underlying causes. Currently, there is little consistency in how a student is evaluated across the different training environments they will encounter in their career, whether in the school, in the field, at home or in simulated training exercises.

 

The Joint ADL Co­Laboratory, in cooperation with the US Navy, the US Army Research, Development, and Engineering Command’s Simulation Training Technology Center, and the US Army’s Program Executive Office for Simulation, Training and Instrumentation, is funding the development of the Learner Assessment Data Model and Authoring Tools (LADMAT) project. This project is focused on development of an assessment data model and associated authoring tools that are capable of capturing complex assessment data across multiple learning and training systems. Specifically, the project will integrate assessment capabilities into a live simulation through the One Tactical Engagement Simulation System and with multiple virtual simulations using the Gamebryo game engine and the Delta3D game engine to ensurethat training objectives are being met by the learner. 

 

One key component of this research effort is to test and evaluate this technology as it is integrated into these dynamic learning environments. This paper will discuss in detail the complexities of assessing performance, the underlying technologies used in this project to simplify the assessment process and how they can be used to help standardize the manner in which the students is assessed throughout their career….

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.

 

 

Mixed Initiative Team Performance Assessment System (MITPAS) For Training and Operation

2007 Paper No. 7398

 

A. Freedy, E. De Visser, G. Weltman

Perceptronics Solutions, Inc.

Sherman Oaks, California

 

Nicole Coeyman

U.S. Army RDECOM-STTC

 

Unmanned vehicles (UVs) are being developed and fielded at an unprecedented rate in various environments for both civilian and military purposes. Despite the term “unmanned,” control of such vehicles requires considerable inputs from those operating the UVs and others in the C2 system.  To optimize such mixed initiative teams we must develop new methodologies and measurements for evaluating and understanding human-robot team performance. The Mixed Initiative Team Performance Assessment System (MITPAS) provides such new methodology. MITPAS consists of models, tools and procedures, including an OneSAF-based simulation environment, with which to measure the performance of mixed manned and unmanned teams in both training and real world operational environments.  This paper describes MITPAS and the results of several initial experiments conducted to validate the measures and gain insight into the effect of robot competence on overall human-robot team performance, operator trust and operator situational awareness.  Our initial results are indicative of the type of new insights into human-robot team behavior that can be gained by combining the measurement power of MITPAS with realistic simulations of tactical UV operations. Consequently, we are working to make the readily-customized MITPAS available to other researchers and developers for use with their simulations, scenarios and special measures.

 

 

This paper is available on the 2007 I/ITSEC CD ROM.

Order the paper from I/ITSEC’s Website.