Decision Analytics, Machine Learning, and Field Experimentation for Defense and Emergency Response

Permanent URI for this collection

https://hdl.handle.net/10125/79163

Browse

Recent Submissions

Now showing 1 - 3 of 3
  • Item
    Navy Tactical Messaging Analytics to Improve Readiness, Planning and Preparedness
    ( 2022-01-04) Dilmaghani, Raheleh ; Turco, Kyle
    There are tens of thousands of official messages exchanged within the Navy’s messaging system daily. The messages carry a tremendous amount of data that is not fully processed nor analyzed. If we could enhance extracting the value of these data, and automate, we could improve speed, readiness, performance, maintenance, outage scheduling, and much more. Although messages follow a prescribed format, it is not uniform across the message types and data entry is not enforced and prone to human interpretation or error in containing all relevant information and correct categorization or labeling. Often there is relevant information in other messages not directly linked or referenced, but if inferred could enhance the observer’s perspective and response to the situation. By analyzing the massive data stored within Navy tactical messages, we could learn about fleet readiness without asking a particular unit or group, “are you ready?” Better yet, we may be able to predict future readiness based on past schedule and performance data if correlated correctly to learn patterns that indicate future mission success. In this paper, we present results from a pilot project where we show the significance of processing available messages for relevant information by not limiting the message types that seem to be the right category or to those with an explicit keyword. This process helps to detect relevant information and create meaning correlations, which could contribute to more informed decision making or to provide the Commander information to make decisions proactively instead of reacting to negative outcomes like casualties or mishaps. Extracting relevant information in turn can contribute to more informed decision making in saving lives, reduced sustainment costs, or repair costs.
  • Item
    Genetic Algorithm Approach for Casualty Processing Schedule
    ( 2022-01-04) Nistor, Marian Sorin ; Pham, Truong Son ; Pickl, Stefan
    Searching for an optimal casualty processing schedule can be considered a key element in the MCI response phase. Genetic algorithm (GA) has been widely applied for solving this problem. In this paper, it is proposed a GA-based optimization model for addressing the casualty processing scheduling problem (CPSP). It aims to develop a GA-based optimization model in which only a part of the chromosome (solution) involves in the evolutionary process. This can result in a less complex training process than previous GA-based approaches. Moreover, the study attempts to investigate two common objectives in CPSP: maximizing the number of survivals and minimizing the makespan. The proposed GA-based model is evaluated on two real-world scenarios in the Republic of Moldova, FIRE, and FLOOD. The paper suggests that GA models with a population size of 500 or smaller can be applied for MCI scenarios. The first objective can help many casualties receiving specialization treatments at hospitals.
  • Item