Thomas Martin Email and Phone Number

Supporting the SharpMinds Army team> Currently developing data cleansing algorithms, including a practical machine learning algorithm that can correct a large proportion of the errors in a data set by mining the data for reliable patterns.> Building predictive modeling techniques into smart input forms to speed input and reduce errors.> Creating data visualizations of the accuracy of the data, the progress of the data cleansing process, and the speed and accuracy of the data cleansing and smart input software.> Inventing algorithms that will cause the data cleansing system to improve over time.

Office of the Deputy Assistant Secretary of Defense (Readiness), Readiness Analytics GroupIn charge of project for the Deputy Assistant Secretary of Defense (Readiness) to develop predictive analytics to anticipate the military readiness of units that are roughly the size of a battalion, ship, or squadron by using measures of the overall personnel quality of the units as well as the usual unit readiness metrics.> The initial stage was to explore the levels of a limited number of personnel quality indicators over the 34-year period FY1980–FY2013. The indicators included high school graduation rates, years of service data, rates of separation for various causes, and a number of others. The study covered all four Armed Services. Each Service was divided into officer and enlisted and then by occupational groups intended to reflect as closely as possible military force elements. For instance, Naval force elements include Surface Warfare, Naval Aviation, Undersea Warfare, Special Warfare/Special Operations, and so on.> The final stage was to develop indices of unit personnel quality that would be useful for predicting unit readiness when combined with other readiness measures. For index development, the tools were the statistical techniques Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA).Produced the Quarterly Readiness Report to the Congress (QRRC), a classified report on the readiness of the U.S. Military (the Armed Services, the Combatant Commands, the Combat Support Agencies, etc.).> Edited and printed the QRRC as part of three-person team.> Implemented new approaches to analyzing and reporting the readiness of the Combatant Commands.> Organized and documented the process of improving the QRRC and all the QRRC-related procedures and documents on the Pentagon SIPR and NIPR networks.

Taught a semester course in college mathematics to Air Force personnel in the 579th Medical Group at Joint Base Anacostia-Bolling

HQ USMC, Programs & Resources Dept, Programs Assessment & Evaluation Dvsn (PA&E)PA&E supports the programming function of HQMC P&R, which is deciding how to allocate resources to meet planning and programming guidance. Much of PA&E’s work is assessing and evaluating the USMC Program Objectives Memorandum (POM).The POM Linear Optimization Tool (PLOT) is software for finding the portfolio of programs that maximizes the total “benefit” the USMC can purchase with given resources. It is PA&E’s principal tool for creating alternate POMs for different toplines and business rules.> Identified software code from the National Institute of Standards and Technology (NIST) that would reduce PLOT run times from about a day down to under an hour, which would have enabled much improved analysis of the Marine Corps’ courses of action for different toplines and business rules.Note. The award of this contract was protested soon after I began work, ending my involvement.

Analyzed and documented the formerly proprietary legacy algorithms that lie at the heart of the software that USMC Manpower and Reserve Affairs (M&RA) has used for over 30 years to plan Marine Corps staffing and each individual Marine’s assignments. The legacy algorithms are highly efficient methods for finding optimal solutions to certain classes of linear programs, such as transportation problems and maximum flow problems. The software included applications to set staffing goals for the enlisted force and the officer force, to classify each basic Marine coming out of boot camp into a training program for an MOS, to develop an MOS classification plan for the new officers coming out of The Basic School (TBS), to assign individual enlisted Marines to individual billets, and to mobilize and assign reserve and retired officers to fill open billets in the event of war.Of course, Marines make the actual M&RA decisions, because these applications are tools, not decision makers.

Advised the Effects-Based Operations Decision Support System (EBODSS) team, which produced the Reconnaissance, Surveillance & Target Acquisition Decision Support System (RSTA-DSS) for the Army Research, Development and Engineering Center (ARDEC) of Picatinny Arsenal. RSTA-DSS is the first software for planning how to deploy combinations of different types of sensing assets, and it can deal with any sensors, (visual, audio, seismic, etc.), platforms (unmanned aerial systems, unmanned ground vehicles, stationary platforms, etc.) and supporting devices (e.g., repeaters) for which specs are available. It shows on-screen what the combined capabilities of the assets are in given terrain (including their fields of view, RF transmission ranges, etc.), and its inference engine recommends combinations of assets for each named area of interest (NAI). The current version runs on a Toughbook so that small units can use it in the field. Applications include planning Army and Marine reconnaissance, surveillance and target acquisition missions, integrated base defense (IBD) and force protection; conducting aspects of intelligence preparation of the battlefield (IPB) and predictive analysis; and training personnel in these areas. > Formulated general model of sensor deployment operations. > Led integrated production team that developed inferencing rules for route reconnaissance. > Created techniques for using RSTA-DSS to plan IBD and force protection. > Developed and implemented computational geometry algorithms in C for NAIs. > Enhanced the “competency of coverage” metric, a measure of how well a given collection of sensors can cover an NAI over the course of a given time period. > Contributed to models of sensor systems and measured sensor fields of view.Developed technical proposals for Army, Navy, Air Force and DARPA solicitations. (Topics and details are EMSolutions proprietary.)

Sample ProjectsHelped develop, test and document the USAF campaign-level model (the Combat Forces Assessment Model, or CFAM), a very large scale mixed integer program. Devised the mathematics and code for the target-accounting submodel that the main model depends on to determine the results over time of campaign operations. Worked on the recent official Verification, Validation & Accreditation (VV&A) of CFAM and its latest implementation, Airstrike 2.0. Made an exhaustive analysis of the mathematical model as implemented in software, re-wrote the CFAM Formulation, and wrote the Verification & Validation report. As a result of the overall V&V effort, the Air Force Analytical Community Steering Group accepted CFAM version 4.0 as the CFAM model of record for the Air Force Standard Analysis Toolkit, and CFAM was officially accredited for use by Dr. Henningsen, Director for Studies and Analyses, Assessments and Lessons Learned, Headquarters U.S. Air Force, Washington, D.C. (The agency is almost always called HQ USAF/A9, or A9 for short.)Designed and implemented the DARPA Director’s Office “Program Transition Database” and its user interface. It was the first comprehensive database of DARPA programs in the history of that agency. Implemented tools for building and retrieving an archive of files (briefings, white papers, etc.) on DARPA programs, as well as designing and coding extensive search features for finding both records in the database and documents in the archive, including full-text search.Created a GPS data post-processor used at Nellis AFB to determine from recorded GPS data when and where military vehicles taking part in wargames moved. (GPS data naturally has random variations that make even stationary vehicles look as if they are hopping from place to place.) Also for ManTech SRS at Nellis AFB, developed an application to read in digital terrain elevation data (DTED) and create perspective renderings of the ground from an aircrew’s point of view.

Taught one of the GWU common core “quantitative reasoning” courses to a class of about 80 students. Not a “quantitative methods” course, it covered George Pólya’s basics of problem solving, with the subject matter drawn mainly from elementary set theory.

Implemented a newly published high efficiency filter bank for the Computer Music Lab in the School of Computer Science in C and rigorously derived the complex frequency responses of the output channels.

Taught at all levels of the undergraduate curriculum, including courses such as first year math for non-majors, linear algebra, advanced calculus, and abstract algebra. Developed software to print cross-sections of a surface on overhead transparencies, which were then mounted one above the other, about an inch apart, to make a three-dimensional graph of the surface. Conducted research in analytic number theory, which uses precisely the same sort of mathematics as signal processing.

Taught calculus, differential equations and techniques of mathematical proof. Conducted research in analytic number theory.