Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Rafe Blaufarb and Liz Clarke, Inhuman Traffick: The International Struggle Against the Transatlantic Slave Trade, a Graphic History
No full text
Reed Ueda, Crosscurrents: Atlantic and Pacific Migration in the Making of a Global America
No full textNeuron-by-neuron mapping of a whole fly brain
No full textThe brain of fruit flies (Drosophila melanogaster) contains 130,000-200,00 neurons, a more approachable number than the 86 billion of the human brain. Last year, the FlyWire research consortium released the first neuron-level adult fruit fly brain map to help understand how the nervous system processes information. Here we describe our effort adding the FlyWire dataset to NeuroMorpho.Org, an open-access database of >260,000 digital tracings of neural morphologies from multiple species. This operation is particularly challenging due to the sheer number (160,427) of FlyWire reconstructions, the morphological irregularities to fix (e.g. long connections and included side branches) and the complex metadata annotation (cell type and anatomical region). NeuroMorpho.Org currently has 36,276 Drosophila cells (13% of the total content, third most represented species) across 17 brain regions. Developmentally, 72% are from adults, 19.4% from larvae, and 8.6% from pupas or unreported stages. The addition of the FlyWire dataset will make Drosophila the species with the most reconstructions in NeuroMorpho.Org, and 97.4% of those tracings will be from adult flies. By utilizing these morphologies, scientists may gain a better understanding of invertebrate brain architectures, which could help reveal how the key organizational principles evolved in more complex organisms such as mammals
SugarNet: Personalized Blood Glucose Forecast for Diabetic Patients with Joint Time-Frequency Learning
No full textDiabetes is a chronic condition characterized by ineffective regulation of blood glucose (BG). Foreseeing upcoming BG trends is crucial to tight glycemic control; however, existing models for BG prediction only use time domain signals. This paper introduces SugarNet, a novel deep learning model for personalized BG forecasts based on food intake, medication, insulin, and BG history. The model augments the inputs with an embedding block aiming for a more expressive representation of the signals. Both the augmented time series and its frequency spectrum are forwarded to two dedicated multi-layer convolution blocks, followed by Long Short-Term Memory (LSTM) networks. The outputs are then fused to generate the forecasts of differences in BG values between current time T and future time T'. Along with novel feature engineering techniques, the model is pre-trained on a set of patients and then fine-tuned and tested on a different set of patients with transferred knowledge. Extensive experiments on 12 type 1 diabetes patients and 100 type 2 diabetes patients yielded improved RMSE by 14.7%-21.8%, demonstrating the superiority of the model over state-of-the-art methods. By providing accurate insights into future BG values, SugarNet can significantly enhance diabetics’ management of their condition
Thermal Tolerance of Japanese Mystery Snails
No full textOne aspect of climate change is an increase in global temperatures, which can impact the range and spread oforganisms. Therefore, studying the heat tolerance of invasive species can help predict their spread under a changingclimate. Japanese Mystery Snails (Heterogen japonica) are invasive freshwater snails from Japan that are highlyadaptable, rapidly reproduce, and lack predators, which are all traits which make them successful invaders. We collectedJapanese Mystery Snails from three different sites in Virginia and Maryland. We then placed 20 individuals and broodsfrom 10 females per site into individual containers in three temperature treatments (35℃,40℃,45℃) and a control on alaboratory bench (25℃) in incubators for seven days. Snails were first acclimated to the treatment temperatures at 1℃/24 hours to prevent thermal shock. We then observed snails daily for mortality for seven days. Results show that themaximum temperature H. japonica can tolerate for a week is between 35℃ and 40℃. At 45℃, there was 100%mortality rate after only 24 hours for both juveniles and adults. At 40℃, there was 100% mortality on the first andsecond day for juveniles and adults, respectively. At 35℃, the snails could survive for the full seven days. Heterogenjaponica’s tolerance is higher temperatures may indicate an advantage and allow for further spread. Their relatively highwater temperature tolerance, along with other aspects of their biology, suggests that expansion into more southernwater bodies in the United States is possible and managers should be alert
Quantifying the Relationship Between Vegetation Characteristics and Avian Diversity in Costa Rica
No full textDeforestation and extreme weather events such as wildfires, floods, and droughts fueled by climate change are causing habitat destruction, posing a severe threat to bird populations worldwide. Currently, over 49% of bird species are in decline, necessitating effective and scalable conservation strategies. This study explores the relationship between vegetation characteristics and bird species diversity, using Costa Rica as a case study. By utilizing satellite-based imagery to assess vegetation health, structure, and texture, along with the novel citizen science dataset eBird to catalog bird species richness and diversity, this work overcomes limitations of scale and cost seen in prior studies. Vegetation indices and bird diversity data were derived for 10km x 10km grid cells over forested areas from 2017 to 2023, followed by an analysis of their association. Results indicate a strong correlation between Normalized Difference Vegetation Index (NDVI), a measure of vegetation health, and both bird species richness and diversity, with NDVI’s spatial mean being the best predictor. Kernel NDVI (kNDVI), a non-linear measure of NDVI, and Enhanced Vegetation Index (EVI), another measure of vegetation health, both emerged as the second-best predictor of both bird species richness and diversity. Additionally, bird species richness was more closely associated with vegetation characteristics than species diversity. Bird species richness proved a more robust diversity measure than diversity derived from observation counts. These findings highlight the importance of vegetation health for bird conservation and can guide targeted conservation efforts and policy decisions
Adaptive Multispectral Object Detection (AMOD) leverages the fusion of RGB and thermal imagery to enhance the performance of object detection models, particularly in varying environmental conditions and distance of the camera from the object. This is part
No full textAdaptive Multispectral Object Detection (AMOD) leverages the fusion of RGB and thermal imagery to enhance the performance of object detection models, particularly in varying environmental conditions and distance of the camera from the object. This is particularly applicable to overhead footage collected by different visual modality cameras on a drone (RGB, thermal etc.). By dynamically adjusting the scaling and transparency between video frames, AMOD can effectively address challenges posed by lighting variations, occlusions, and thermal inconsistencies. The fusion process integrates the complementary strengths of RGB and thermal data: RGB provides high-resolution texture and color information, while thermal imagery offers robust detection in low-light and obscured environments. This research investigates the impact of varying scaling and transparency on object detection model performance within this multispectral framework. By modifying the transparency levels, the model can emphasize either RGB or thermal data based on the environmental context, enhancing feature extraction and object localization. Scaling adjustments enable the alignment of features across different resolutions, ensuring coherent fusion and reducing artifacts. This research highlights the potential of AMOD in applications requiring reliable object detection across varied operational contexts, such as military surveillance, search and rescue, and autonomous navigation
ReqFusion: Use of Natural Language Processing to Automate the Integration of Legacy and New Requirements for the United States Air Operations Command.
No full textRequirements dictate funding in the United States Department of Defense (DoD); some of these older “legacy requirements” were written up to 40 years ago. There are thousands of pre-existing legacy requirements in the DoD and new requirements are still being created every year. Sorting through the legacy requirements to find out if one already exists is a tedious task. We created a tool for the Air Operations Command (AOC) to integrate the new and legacy requirements to output combined requirements and allow the AOC to receive funding for their program plans. Natural Language Processing (NLP) is a field of Artificial Intelligence (AI) that allows computers to understand, interpret, and generate human language. We used various pre-trained NLP models and the Groq API to develop an NLP system that combines similar requirements. We created ReqFusion which is a web application that leverages this NLP system. Users can upload new requirements to the web app, which are integrated with pre-existing legacy requirements provided to us by the DoD. Integrated requirements are outputted into a CSV file for compatibility and as a data table displayed on the web app. ReqFusion can optimize determining and allocating funding in the DoD so that programs can be carried out and capabilities can be delivered sooner
Applications of Least Squares to find optimal solutions to real world problems involving circle regression.
No full textThe Least Squares Method is a mathematical technique that helps to estimate parameters using a regression analysis approach based on minimizing residuals, which is the difference between data (an observed value) and a computed value. Most popular estimations include linear, quadratic, polynomial fits to approximate the given data. The best fit to the data using these functions are determined through the respective correlation coefficients. However, there are applications when all these well-known fits do not give good approximations, such as when the data is in the form of a circle; in which case, a different regression approach is required. Real-world data that arise from applications don’t necessarily exhibit trends that can be captured by standard regression algorithms, for example predicting the diameter of the head of a baby through ultrasound data to predicting optimal location of an aircraft from satellite stations at different locations. In this work, one formulates this circle regression problem using a matrix approach to find the optimal solution. Using a matrix approach, we expand the circle regression approach to finding the most accurate location of an aircraft. Our results suggest that the proposed method is reliable and robust and can be used for other related real world applications