USMA Digital Commons (United States Military Academy, West Point)
Not a member yet
1355 research outputs found
Sort by
Aquaticus:Publicly Available Datasets from a Marine Human-robot Teaming Testbed
No full textIn this paper, we introduce publicly available human-robot teaming datasets captured during the summer 2018 season using our Aquaticus testbed. Our Aquaticus testbed is designed to examine the interactions between human-human and human-robot teammates while situated in the marine environment in their own vehicles. In particular, we assess these interactions while humans and fully autonomous robots play a competitive game of capture the flag on the water. Our testbed is unique in that the humans are situated in the field with their fully autonomous robot teammates in vehicles that have similar dynamics. Having a competition on the water reduces the safety concerns and cost of performing similar experiments in the air or on the ground. By having the competitions on the water, we create a complex, dynamic, and partially observable view of the world for participants while in their motorized kayak. The main modality for teammate interaction is audio to better simulate the experience of real-world tactical situations - ie fighter pilots talking to each other over radios. We have released our complete datasets publicly so that we can enable researchers throughout the HRI community that do not have access to such a testbed and may have expertise other than our own to leverage our datasets to perform their own analysis and contribute to the HRI community
Predictive Mathematical Models of Weight Loss
No full textPurpose of Review
Validated thermodynamic energy balance models that predict weight change are ever more in use today. Delivery of model predictions using web-based applets and/or smart phones has transformed these models into viable clinical tools. Here, we provide the general framework for thermodynamic energy balance model derivation and highlight differences between thermodynamic energy balance models using four representatives. Recent Findings
Energy balance models have been used to successfully improve dietary adherence, estimate the magnitude of food waste, and predict dropout from clinical weight loss trials. They are also being used to generate hypotheses in nutrition experiments. Summary
Applications of thermodynamic energy balance weight change prediction models range from clinical applications to modify behavior to deriving epidemiological conclusions. Novel future applications involve using these models to design experiments and provide support for treatment recommendation
Deep Learning for Inexpensive Image Classification of Wildlife on the Raspberry Pi
No full textAnimal conservationists need unobtrusive methods of observing and studying wildlife in remote areas. Many commercial options for wildlife observation are expensive, obtrusive, or sub-optimal in remote environments. In this paper, we explore the viability of a Raspberry Pi-based camera system augmented with a deep learning image recognition model for detecting wildlife of interest. Unlike traditional sensor nodes that would have to transmit every captured image, localized image recognition enables only pictures of desired animals to be transferred to the user. For the purposes of this study, we use TensorFlow and Keras to create a convolutional neural network that runs on a Raspberry Pi 3B+. We trained the model on nearly 3,600 images gathered from publicly available image databases that are split into three classes. Our experiments suggest that our system can detect snow leopards with between 74 percent and 97 percent accuracy. We believe that our results show the viability of employing deep learning image recognition models on the Raspberry Pi to create an inexpensive system to observe wildlife
Intelligent Feature Engineering for Cybersecurity
Get PDFFeature engineering and selection is a critical step in the implementation of any machine learning system. In application areas such as intrusion detection for cybersecurity, this task is made more complicated by the diverse data types and ranges presented in both raw data packets and derived data fields. Additionally, the time and context specific nature of the data requires domain expertise to properly engineer the features while minimizing any potential information loss. Many previous efforts in this area naively apply techniques for feature engineering that are successful in image recognition applications. In this work, we use network packet dataflows from the Defense Research and Engineering Network (DREN) and the Engineer Research and Development Center\u27s (ERDC) high performance computing systems to experimentally analyze various methods of feature engineering. The results of this research provide insight on the suitability of the features for machine learning based cybersecurity applications
Cyber Threat Report 04 March 2019
Get PDFArmy Cyber Institute Cyber Threat Report
Tech Trends: Stories and Highlights QuadrigaCX cryptocurrency exchange customers may be out $190M after founder dies with passwords Stone Panda Hacks Norwegian and US companies Supply Chain Attacks Spiked 78% in 2018 Congress considering exchange program for federal, industry, and academia cyber expert
Allometric scaling of weight to height and resulting body mass index thresholds in two Asian populations.
Get PDFBACKGROUND: Body mass index (BMI) represents a normalization of weight to height and is used to classify adiposity. While the capacity of BMI as an adiposity index has been experimentally validated in Caucasians, but there has been little testing Asian populations.
METHODS: To determine whether weight scales to height squared in Asian Indians across the general population and in Asian Indian tribes an allometric analysis on the power law model, W = αH
RESULTS: The unadjusted power was β = 2.08 (s = 0.02). The power for the general population (non-tribal) was β = 2.11 (s = 0.02). Powers when adjusted for tribe ranged from 1.87 to 2.35 with 24 of the 33 tribes resulting in statistically significant (p \u3c 0.05) differences in powers from the general population. The coefficients of the adjusted terms ranged from -0.22 to 0.26 and therefore the scaling exponent does not deviate far from 2. Thresholds for BMI classification of overweight in the KNHANES database were BMI = 21 kg/m
CONCLUSIONS: Our study confirms that weight scales to height squared in Asian Indian males even after adjusting for tribe membership. We also demonstrate that optimal BMI thresholds are lower in a Korean population in comparison to currently used BMI thresholds. These results support the application of BMI in Asian populations with potentially lower thresholds
Simulation of Man in the Middle Attack On Smart Grid Testbed
Get PDFOver the past decade, the frequency of cyber attacks against power grids has steadily increased, requiring researchers to find and patch vulnerabilities before they can be exploited. Our research introduces the prototype of a man-in-the-middle attack to be implemented on a microgrid emulator of a smart grid. We present a method of violating the integrity and authentication of packets that are using the IEEE Synchrophasor Protocol in a controlled environment, but this same approach could be used on any other protocol that lacks the proper overhead to ensure the integrity and authenticity of packets. In future research, we plan to implement and test the attack on the previously mentioned smart grid testbed in order to assess the attacks feasibility and tangible effects on Wide Area Monitoring and Control applications, as well as propose possible countermeasures. For this paper, we developed a working simulation of our intended attack using the software ModelSim 10.4. The attack will modify network packet data coming from a Schweitzer Engineering Labs (SEL) Phasor Measurement Unit (PMU) hardware sensor, which provides a stream of precise timing values associated with current and voltage values, as these measured values are en route to the Open Phasor Data Concentrator (OpenPDC) application running on a Windows server. Our simulation provides and validates all of the necessary code in order to program a Field Programmable Gate Array and execute our attack on the testbed in future research
A Machine Learning Framework for Building Passive Surveillance Photogrammetry Models
No full textDetermining the geographic location of an object using two-dimensional (2D) images recorded at high-oblique angles is a nontrivial problem. Existing methods to solve this problem rely on parameters that are either difficult to measure or are based on assumptions. This paper investigates the accuracy of building photogrammetric models using machine learning. Our novel approach involves the collection of training examples before using supervised learning to build a nonlinear, multitarget prediction model. We collected training examples using an unmanned ground vehicle (UGV) that moved throughout the fields of view of multiple cameras. The UGV was tracked and bounded using existing computer vision techniques. With each image frame, the center pixel position (x, y image coordinates) of the vehicle and its bounding box area (in pixels) were mapped to its current GPS coordinates. Multiple machine learning models were created using various combinations of cameras to determine the key features for building accurate photogrammetric models. Data was collected under realistic conditions for ground-based surveillance systems, which may require cameras to be placed at low elevations and high-oblique angles. We found the prediction accuracy of our models to be between 0.58 and 3.54 meters depending upon a number of factors, including the locations, heights, and orientations of the cameras used
Cyber Threat Report June 07, 2019
Get PDFArmy Cyber Institute Cyber Threat Report
Tech Trends: Stories and Highlights Huawei summary and what the future holds China repurposes NSA Hacking Tools Russian hackers breach three U.S. anti-virus companies UK Government shares intel on malicious activity with 16 NATO Allies Amazon debuts delivery drone and claims to be operational in month