Charles River Analytics, a developer of intelligent systems solutions, received a follow-on contract to continue development of its Model Analyst’s Toolkit (MAT). MAT is a software application to help scientists and analysts create, analyze, refine, and validate complex scientific models using real-world data. MAT’s initial development focused on developing and validating scientific models of socio-political behavior. It is now being extended to other domains, supported by a contract awarded by the Office of Naval Research (ONR) for over $900,000.
Dr. Scott Neal Reilly, Vice President, Decision Management Systems, said, “MAT provides a deep understanding of both data and models, creating better explanations of why a model does what it does, uncovering invalid model assumptions or results, comparing models against large data sets and using that data to improve the models, and providing greater trust of validated models. It is a potent tool for scientists in many fields who are attempting to understand complex and interlinked behaviors in a systematic way.”
MAT was originally developed under an effort for ONR to validate human social cultural behavior (HSCB) models. For that effort, MAT supported social scientists in visualizing and exploring data, developing causal models, and validating those models against available data. The figure here shows some of MAT’s tools for visualizing data and developing and validating models.
A number of MAT’s features are displayed in the screenshots shown here.
In this follow-on effort, Charles River will extend MAT’s support for model development, analysis, refinement, and validation. It will also enhance MAT to analyze and repair data, extend MAT to use machine learning approaches to automatically suggesting improvements to models, and extend MAT to work across multiple domains, from soft-science models of HSCB to hard-science models of neurophysiological phenomena.
Dr. Neal Reilly continued, “The MAT tool has been very successful in helping users in the HSCB community to develop better models. We believe the current effort will not only significantly improve on these capabilities, but will demonstrate that many of these same approaches are useful in a variety of other scientific fields, such as neuroscience.”
Charles River is also conducting neurophysiological studies in other efforts, including SONNET for the Defense Advanced Research Projects Agency’s (DARPA) Narrative Networks (N2) program.
To learn more about the features and benefits, visit the MAT page.