USMA Digital Commons (United States Military Academy, West Point)
Not a member yet
1355 research outputs found
Sort by
Deep Vision for Breast Cancer Classification and Segmentation
No full text(1) Background: Female breast cancer diagnoses odds have increased from 11:1 in 1975 to 8:1 today. Mammography false positive rates (FPR) are associated with overdiagnoses and overtreatment, while false negative rates (FNR) increase morbidity and mortality. (2) Methods: Deep vision supervised learning classifies 299 × 299 pixel de-noised mammography images as negative or non-negative using models built on 55,890 pre-processed training images and applied to 15,364 unseen test images. A small image representation from the fitted training model is returned to evaluate the portion of the loss function gradient with respect to the image that maximizes the classification probability. This gradient is then re-mapped back to the original images, highlighting the areas of the original image that are most influential for classification (perhaps masses or boundary areas). (3) Results: initial classification results were 97% accurate, 99% specific, and 83% sensitive. Gradient techniques for unsupervised region of interest mapping identified areas most associated with the classification results clearly on positive mammograms and might be used to support clinician analysis. (4) Conclusions: deep vision techniques hold promise for addressing the overdiagnoses and treatment, underdiagnoses, and automated region of interest identification on mammography
Improving Mission Assurance Assessments for Resilience of Military Installations
No full textIt is critical to improve the resilience of military installations and their complex infrastructure systems to strengthen response to the uncertainty and threat driven by the increasing frequency and severity of natural and human-made disasters. This research addresses a considerable gap in the existing Department of Defense (DOD) Mission Assurance Framework between the infrastructure assessment process and resilience considerations, and integrates a resilience matrix that converts qualitative assessment data into a quantifiable and interactive resilience decision support tool. The integration of the resilience matrix provides a quantitative visual tool to communicate the impact of decisions made using the tradespace analysis. This methodology provides a framework to improve the selection of projects that enhance the resilience of military infrastructure systems and assist decision makers in understanding how a single project may influence the resilience of multiple systems. The results of this research, which were built for a specific installation, are broadly applicable and can support engineers in the design and/or management of infrastructure systems to improve resilience in an efficient manner
Combining Wargaming with Modeling and Simulation to Project Future Technology Requirements
No full textThe rapid growth and widespread availability of technology has allowed enemies to dynamically develop countermeasures to military systems. Therefore, it is imperative that military systems be designed to account for these countermeasures. As such, technology roadmapping should be a critical activity in the acquisition of defense systems. Technology roadmaps provide a strategic vision for a system that accounts for the operational context, including evolving needs and technology changes. However, the operational context can be difficult to predict. This article suggests using wargaming coupled with combat simulation to better understand the operational context to allow for testing and refining technology roadmaps. Wargaming requires teams to roleplay friendly and enemy units to determine how each side adapts with the implementation of a new military system. Computer-based simulations can then convert the qualitative results from the wargame into quantitative metrics that further inform the roadmap. A case study is presented for a technology roadmap associated with an armored exoskeleton. Wargaming forecasted the countermeasures implemented by the enemy and the associated responses. The wargame results were coupled with models to quantitatively forecast the change in the warfighter\u27s survivability and lethality. The wargame was then used to inform the technology roadmap
Equipment Turn-In and Transfer Process Modeling
No full textIn military bases across the country, there is excess equipment that costs money to store while it is going unused. When this excess grows large enough, there is an obvious need to turn it in or transfer it to a location where it will be used, but there is not currently a process in the military to handle large stocks like Fort Hood, Texas. This research examines the possibilities of what this turn-in process can look like and how to create a modernized and synchronized method that can be employed across the Army. Modeling this proposed process in ProModel should highlight issues with the current flow and recommend system improvements. Although the process has just been implemented at Fort Hood, there are several system improvements that have been identified by backlogs in the discrete-event simulation including reducing process times, allowing workers to perform tasks in parallel, and reducing time needed to order and install parts. These improvements will make the equipment turn-in process more manageable for Army-wide implementation and success
Dive into Systems: A Free, Online Textbook for Introducing Computer Systems
No full textThis paper presents our experiences, motivations, and goals for developing Dive into Systems [17], a new, free, online textbook that introduces computer systems, computer organization, and parallel computing. Our book\u27s topic coverage is designed to give readers a gentle and broad introduction to these important topics. It teaches the fundamentals of computer systems and architecture, introduces skills for writing efficient programs, and provides necessary background to prepare students for advanced study in computer systems topics. Our book assumes only a CS1 background of the reader and is designed to be useful to a range of courses as a primary textbook for courses that introduce computer systems topics or as an auxiliary textbook to provide systems background in other courses. Results of an evaluation from students and faculty at 18 institutions who used a beta release of our book show overwhelmingly strong support for its coverage of computer systems topics, its readability, and its availability. Chapters are reviewed and edited by external volunteers from the CS education community. Their feedback, as well as that of student and faculty users, is continuously incorporated into its online content. We anticipate releasing version 1.0 of the book in spring of 2021, and a release candidate is currently available at https://diveintosystems.org
Machine Learning for Raw Network Traffic Detection
No full textIncreasingly cyber-attacks are sophisticated and occur rapidly, necessitating the use of machine learning techniques for detection at machine speed. However, the use of machine learning techniques in cyber security requires the extraction of features from the raw network traffic. Thus, subject matter expertise is essential to analyze the network traffic and extract optimum features to detect a cyber-attack. Consequently, we propose a novel machine learning algorithm for malicious network traffic detection using only the bytes of the raw network traffic. The feature vector in our machine learning method is a structure containing the headers and a variable number of payload bytes. We propose a 1D-Convolutional Neural Network (1D-CNN) and Feed Forward Network for detection of malicious packets using raw network bytes
Data-driven CFD Scaling of Bioinspired Mars Flight Vehicles for Hover
No full textOne way to improve our model of Mars is through aerial sampling and surveillance, which could provide information to augment the observations made by ground-based exploration and satellite imagery. Flight in the challenging ultra-low-density Martian environment can be achieved with properly scaled bioinspired flapping wing vehicle configurations that utilize the same high lift producing mechanisms that are employed by insects on Earth. Through dynamic scaling of wings and kinematics, we investigate the ability to generate solutions for a broad range of flapping wing flight vehicles with masses ranging from insects O(10−3) kg to the Mars helicopter Ingenuity O(100) kg. A scaling method based on a neural-network trained on 3D Navier-Stokes solutions is proposed to determine approximate wing size and kinematic values that generate bioinspired hover solutions. We demonstrate that a family of solutions exists for designs that range from 1 to 1000 g, which are verified and examined using a 3D Navier-Stokes solver. Our results reveal that unsteady lift enhancement mechanisms, such as delayed stall and rotational lift, are present in the bioinspired solutions for the scaled vehicles hovering in Martian conditions. These hovering vehicles exhibit payloads of up to 1 kg and flight times on the order of 100 min when considering the respective limiting cases of the vehicle mass being comprised entirely of payload or entirely of a battery and neglecting any transmission inefficiencies. This method can help to develop a range of Martian flying vehicle designs with mission viable payloads, range, and endurance
Drone Assisted Targeting for Direct Fire Engagements
No full textThe goal of the project is to develop a drone and recovery station system capable of tracking a target and round fired from a tank. Key aspects of the project include maintaining a level platform through the shot, performing accurate localization, and reacting to movements in the tank turret. The goal is for the design to be able to operate in GPS-denied and light-denied environments. After significant testing and research, the drone will be placed 10 meters behind and 20 meters above the tank muzzle and the recovery station will be placed on the rear of the tank. The drone will perform localization using visual-inertial odometry (VIO) and will react to movements in the tank turret by using a filtered IMU to send up the tank’s azimuth. All communication will be transmitted over WiFi and the video from the camera will be livestreamed on the recovery station laptop. In this semester, key areas of focus will be validating the VIO algorithm, developing a method for aligning the drone with the tank turret, establishing communication over WiFi, and then conducting final testing