The last few years have marked immense growth in the development of digital twins as developers continue to devise strategies to ensure their devices replicate their physical twin’s actions in a real-time virtual environment. The complexity and predictability of these environments can be the deciding factor for adequately testing a digital twin. As of the last year, a digital twin was in development for a capstone project at Arizona State University: CIA Research Labs - Mechanical Systems in Virtual Environments. The virtual device was initially designed for a fixed environment with known ahead-of-time obstacles. Due to the fact that the device was expected only to be traversing set environments, it was unknown how it would handle being driven in an environment with more randomized and unexpected obstacles. For this paper, the device was test driven in the original and environments with various levels of randomization to see how usable and durable the digital twin is despite only being built for environments with expected object locations. The research allowed the creators of this digital twin, utilizing the results of the trial runs and the number of obstacles unsuccessfully avoided, to understand how reliable the controls of the digital twin are when only trained for fixed terrains

2018, Google researchers published the BERT (Bidirectional Encoder Representations from Transformers) model, which has since served as a starting point for hundreds of NLP (Natural Language Processing) related experiments and other derivative models. BERT was trained on masked-language modelling (sentence prediction) but its capabilities extend to more common NLP tasks, such as language inference and text classification. Naralytics is a company that seeks to use natural language in order to be able to categorize users who create text into multiple categories – which is a modified version of classification. However, the text that Naralytics seeks to pull from exceed the maximum token length of 512 tokens that BERT supports – so this report discusses the research towards multiple BERT derivatives that seek to address this problem – and then implements a solution that addresses the multiple concerns that are attached to this kind of model.

This thesis explores how large scale cyber exercises work in the 21st century, going in-depth on Exercise Cyber Shield, the Department of Defense’s largest unclassified cyber defense exercise run by the Army National Guard. It highlights why these cyber exercises are so relevant, going over several large scale cyber attacks that have occurred in the past year and the impact they caused. This research aims to illuminate the intricacies around cyber exercise assessment involving manual vs automated scoring systems; this is brought back to work on creating an automated scoring engine for Exercise Cyber Shield. This thesis provides an inside look behind the scenes of the operations of the largest unclassified cyber defense exercise in the United States, including conversations with the Exercise Officer-In-Charge of Cyber Shield as well as a cyber exercise expert working on assessment of Exercise Cyber Shield, and the research also includes information from past final reports for Cyber Shield. Issues that these large scale cyber exercises have faced over the years are brought to light, and attempts at solutions are discussed.