12466 research outputs found
Sort by
Gustav Klemp World War I Narrative: German Translation
No full textGerman language text translation from a manuscript originally handwritten in German. This manuscript is a narrative that documents the military service experience of Gustav Klemp, a Prussian medic with the German army who served on the Eastern Front during World War I.German language text translation from a manuscript originally handwritten in German. This manuscript is a narrative that documents the military service experience of Gustav Klemp, a Prussian medic with the German army who served on the Eastern Front during World War I
Hybrid Deep Learning for Intelligent Chemical Sensing
Full text linkVolatile organic compounds (VOCs) are fundamental elements of both the global atmosphere and aquatic ecosystems, acting as infochemicals that enable inter- and intra- species communication. The identification of VOCs holds great promise in various fields including food engineering, environmental monitoring, and medical diagnosis. However, the sensitivity of gaseous VOCs to temperature fluctuations, stemming from their low molecular mass and high vapor pressure, presents a hurdle to the widespread adoption of Electronic Noses (E-noses). Additionally, there exists a research gap in the development of portable devices capable of detecting liquid VOCs in solution. Moreover, neural networks such as CNN and Transformer have not been extensively explored in the field of VOC classification. To address these issues, this dissertation endeavors to improve sensor analysis systems for VOCs in both gas and liquid states through the utilization of hybrid deep learning methods.The first section of the dissertation concentrates on differentiating the gaseous profiles of VOCs originating from fruit scents, detected by metal oxide sensors. An intelligent gas sensing system (E-nose) is designed and fine-tuned, utilizing Piecewise Equation Fitting (PEF), Decision Tree, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and One-dimension convolutional neural network (1D-CNN). The convolutional layers in 1D-CNN showcased superior feature extraction capability compared to conventional methods, enabling precise identification of digital scents and monitoring of the decomposition dynamics of various foods. To overcome the limited adoption of E-nose technology, the second section of the dissertation is to improve its robustness. A dataset was compiled based on the VOCs emitted by vegetarian foods at different temperatures. A fusion of 1D-CNN and Transformer Encoder, trained on this dataset, significantly improved the gas sensing system’s robustness and generalization capabilities. The Transformer leverages global feature interactions to enrich the set of temperature-independent features. Furthermore, feature selection within the fully-connected layers proved crucial in isolating temperature-independent features from complex temperature-variant datasets, thereby enhancing the model's robustness to temperature fluctuations. In the third section of the dissertation, an innovative electrochemical sensing approach utilizing cyclic voltammetry (CV) with an electrolyte based on ionic liquid was developed. The electrolyte enables selective reactions with VOCs, resulting in distinct voltammograms. Building upon these voltammograms, the proposed 1D-CNN significantly enables precise and robust classification of various liquid VOC analytes. Additionally, for quantification, the system effectively categorized methanol volumes ranging from 0 µL to 50 µL in 10-µL increments, achieving exceptional accuracy. The final section summarizes the contributions of 1D convolutional layers, fully connecting layers and Transformer encoders in synergistically elevating the precision and robustness of E-nose technology and the innovative electrochemical VOC detector. Furthermore, the section proposes future research directions to facilitate the practical implementation and widespread adoption of these two systems. This dissertation offers comprehensive discrimination, quantification, and monitoring of VOCs in both gas and liquid states using machine learning methods. The approaches centered on 1D-CNN exhibit superior performance in accuracy, generalization, and robustness. Such promising technologies hold significance for the advancement of sensor device technology in the future, and are highly attractive for the identification and analysis of VOCs
Military Clashes in North Darfur, April 6-18 2024
Full text linkThis report summarizes the large-scale ethnic-based violence in North Darfur, Sudan, between April 6-18, 2024. In early April, a local Arab militia led by the Rapid Support Forces (RSF) allegedly sought retribution for stolen cattle.Produced with the support of the Bureau of Conflict and Stabilization Operations, United States Department of State
Investigating the Effects of Increasing Sea Surface Temperature (SST) in the Northwest Atlantic Ocean on Tornado Activity in the Northeast United States 1990 - 2022
Full text linkThere are global effects caused by a warming climate that will continue to cause severe impacts to the planet. A warming climate has been causing some of the highest increases globally to Sea Surface Temperature (SST) in the Northwest Atlantic (Pershing et al., 2015). These rising ocean temperatures allow for more moisture from the ocean to evaporate into the atmosphere creating instability in the atmosphere and providing the energy that is favorable for storm development. In recent years, the Northeast United States has been impacted by hurricanes, significant rain and flooding events, and tornado activity. The Northeast U.S. is experiencing an upward trend in environments favorable for tornado development because of potential increases to instability parameters such as Convective Available Potential Energy (CAPE) (Brooks et al., 2014). In addition, a Global Climate Model using variables such as varying levels of geopotential heights and 850mb level temperatures suggest that the eastern Mid-Atlantic is again projected to be the region at greatest risk for an increase in the number of tornado days that have tornadoes with an intensity of (F2+) for years 2050 and 2090 (Lee 2011). There is sufficient evidence suggesting that the Northeast region is at risk of increased tornado threats and proves the importance of identifying areas in the Northeast United States that have been experiencing increasing trends in tornado activity, and why it is essential to identify variables that are most influential to annual tornado activity. These extreme weather events put a large population and infrastructure at risk of complete destruction, with severe consequences to the economic stability of the region. This study analyzed tornadic trends between two time periods (1990 to 2004) and (2005 to 2022) using spatio-temporal analysis with GIS for the Northeast U.S. by analyzing the NOAA point tornado dataset through aggregation of touchdown locations both spatially and temporally into 50km hexbins. Guo et al., (2016) suggested for future studies to grid tornado data at different resolutions to further evaluate spatially regionalized tornado variability in the U.S. This is why this study examined data over smaller spatial extents to define localized areas that are experiencing the greatest impacts due to a changing environment. This study also examined annual tornadic activity for the months April to November in the Northeast U.S. for the years (1990 to 2022) and the relationship of tornadic instability parameters, large scale atmospheric patterns, and SSTs in the Northwest Atlantic. A statistical approach was used to determine what predictor variables are most efficient at predicting annual tornado activity and for examining years that outbreaks occurred. Research was also conducted on whether there was an increase in the number of more intense tornadoes between the two time periods and if specific variables impact intensity. Major findings of this study are: statistically significant positive trends in tornado counts in the Northeast U.S. are occurring near coastal areas of the Northwest Atlantic, the SSTs for the Northwest Atlantic and Storm Relative Helicity (SRH) are statistically significant predictor variables and account for 13% explained deviance for annual tornado counts. SRH is the measure of the potential for cyclonic updraft rotation in right-moving supercells, and it was the only statistically significant predictor variable in predicting years with outbreaks/no outbreaks and contributed to the model having an explained deviance of 40%. SRH, NW SST’s and the North Atlantic Oscillation (NAO) were all found to be statistically significant explanatory variables for tornadoes rated as F2 or higher during outbreaks. These findings indicate that SSTs have statistical significance and suggests that warming waters do have certain influences on tornado frequency, intensity, and location of tornado activity within the Northeast United States. If SSTs continue to rise, populations closest to the coast could be most susceptible to future extreme weather events. Potential research with this topic could focus on some of the statistically significant predictor variables that were identified in relation to tornadoes rated (F2+) and topographic features within the Northeast to examine how the terrain influences the locations of the more severe storms
Ground-based Light Curve Follow-up Validation observations of TESS object of interest TOI 3521.01
No full text“The NASA Transiting Exoplanet Survey Satellite (TESS) has identified thousands of exoplanet candidates, including TOI-3521.01, a potential exoplanet orbiting a star approximately 762 parsecs away from Earth. This paper presents ground-based follow-up observations of TOI-3521.01 using the George Mason University 0.8m telescope. Through data reduction and multi-aperture photometry with AstroImageJ, we produced a light curve that indicates a possible transit event. Despite significant noise and challenges in data collection, the observed transit depth and duration align closely with predictions from TESS data. Although the light curve suggests the existence of TOI-3521.01, further analysis, including radial velocity measurements and absorption spectroscopy, is necessary to confirm the exoplanet and better understand its characteristics. We also discuss the limitations of our observations and recommend additional follow-up studies.
Ground-based Light Curve Follow-up Validation observations of TESS object of interest TOI 5868.01
Full text link“This study aimed to provide additional confirmation, details, and classification for TESS Object of Interest (TOI) 5868.01. After obtaining the observation data of TOI 5868.01 from George Mason University, we could generate a stellar light curve of this object using AstroImageJ. Upon producing this light curve, it is unclear if a transit is present. The data had an initial scatter percentage (RMS) of 2.263%. We discovered that by using detrending parameters like AIRMASS and Width_T1, the RMS dropped to 1.976%. This is still not perfect though. As a result, while suggestive, the study's results are not definitive. They suggest that the start might have had a transit, however due to the noise in the data, further analysis is recommended. There does seem to be some activity in between the ingress and egress, but the graph only starts to fluctuate or dip much after the predicted ingress has started. Though a definitive conclusion cannot be made just yet, additional information needs to be gathered and compared with the available data to validate TOI 5868.01 as a transit.
ENCODING STRATEGIES, HEMIFIELDS, AND SHAPE COMPLEXITY, OH MY! AN INVESTIGATION OF METHODS TO IMPROVE MEMORY PERFORMANCE
Full text linkWorking memory (WM) is a form of temporary memory where information is stored for brief periods of time and is manipulated to support on-going cognitive tasks, such as counting or performing mathematical computations. WM is limited by the number of items that may be maintained, which is referred to as capacity. In order to overcome limits in WM capacity, researchers have investigated many methods to overcome these limitations and improve performance. The present research conducted a series of experiments to understand the effects of encoding strategies on WM capacity measures, and examined the potential for multiple strategies to synergize to improve WM performance. In this series of experiments, encoding strategies came in the form of task instructions. Participants were either provided instructions to remember the locations of items (which served as the as the control condition) or were provided instructions to remember the locations of items by forming patterns between the items, like a constellation (referred to as constellation instructions and served as the experimental condition). Study 1 investigated the effects of encoding strategies on spatial WM performance and if it could be indexed by the same neurophysiological markers of visual WM capacity. The results of Study 1 found that spatial WM could be indexed by neurophysiological markers of visual WM and for participants in the constellation condition, the neurophysiological marker supported the notion that participants formed connections between locations. Additionally, there were differences in the duration of the neurophysiological across instruction conditions where participants in the constellation condition had prolonged activity than participants in the control condition. Study 1 did not see effects in capacity performance but did see differences reflected in neurophysiological markers of spatial WM, providing evidence that participants were indeed forming a constellation when instructed to do so. Study 2 investigated the effects of encoding strategies with task instructions combined with the effects of stimulus organization to determine if there were additive effects of these two strategies. Stimulus organization was varied by presenting items within a single visual hemifield, thus utilizing a single hemisphere of the brain, or presented items across both visual hemifields, utilizing both hemispheres of the brain. The results of Study 2 found that participants recalled more items when items when stimuli were arranged bilaterally (across visual hemifields) than unilaterally (within a visual hemifield). There was not an effect of task instruction on capacity measures; however, there were differences observed in reaction time measures. Participants in the constellation condition responded faster than participants in the control condition. While Study 2 found different effects of stimulus organization and encoding strategies on spatial WM measures, it is not possible to understand the mechanisms behind these effects. To further investigate the role of spatial organization of stimulus arrays and encoding strategies, Study 3 introduced an additional factor of stimulus complexity. The goal of this manipulation was to determine if WM capacity improvements were due to an increase in the amount of resources available (i.e., each hemisphere has its own reserve of resources) or if WM capacity improvements were due to an increased number of “slots” available to encode and store items (i.e., each hemisphere has its own number of slots). Stimulus complexity was varied by types of shapes: complex shapes in the form of irregular polygons or simple shapes in the form of squares. If memory capacity did not vary across stimulus complexity, it would support the notion that there are more slots available; however, if memory capacity did vary with stimulus complexity it would support the notion there are more resources available. The results of Study 3 indicated partial support for greater resource availability as there were differences in performance with shape complexity; however, these results were only seen in the constellation condition. Overall, these studies find that there are improvements in spatial WM performance when participants are provided encoding strategies which can be further improved when information is organized in a way that leverages both hemispheres of the brain
Ground-based Light Curve Follow-up Validation Observations of TESS Object of Interest TOI 5612.01
Full text link“Several studies in astronomy have focused on exoplanet analysis; however, many exoplanet candidates still remain to be confirmed. This study aims to discover if the Transiting Exoplanet Survey Satellite (TESS) Object of Interest 5612.01 is an exoplanet transit or if it is a false positive caused by an event such as a near eclipsing binary. This analysis ensures the accuracy of the transit discovered by TESS as it provides a secondary source of confirmation for the data discovered by TESS as well as deeper analysis on specific aspects of the transit such as if other stars may have affected the transit due to them experiencing separate astrological events. Using observations from George Mason University’s observatory as well as AstroImageJ and NASA reference materials, this study generated plate-solved images, Near Eclipsing Binary (NEB) analysis, and a light curve. Using the statistical methods of root mean squared analysis and the bayesian information criterion, a model of the transit was generated through this data. However, no result was found due to error in the data set, likely caused by cloud cover or light pollution. This paper provides a demonstration of the flaws of the data set, showing that TOI 5612.01 requires reanalysis and offers commentary on conditions that should be avoided for future analysis of this exoplanet.
TRUSTED SOURCES AS MYTHBUSTERS: EVALUATING CLIMATE MYTH DEBUNKING ON SOCIAL MEDIA BY TRUSTED CLIMATE INFORMATION SOURCES
Full text linkPrevious research has shown that climate scientists, broadcast meteorologists, and primary care providers are trusted sources for climate information. The present study investigates whether these trusted sources can serve as “mythbusters”—professionals who can debunk misinformation as part of their climate change communication efforts—while also investigating the roles of source trust and alignment of a debunking topic with the source’s specific field of expertise. In a survey-based experiment (n = 900; 49.7% female, 48.75 male; 76.1% White/Caucasian, 12.8% Black/African American, 6.2% Asian/Pacific Islander, 3.6% Hispanic/Latino, median age 46 years), participants were randomly shown one of ten possible mock Facebook posts in a 3 sources (broadcast meteorologist, climate scientist, doctor) x 3 myths (cold weather stops with climate change, there is no consensus on the cause of climate change, climate change is good for human health) design, plus a control condition. Three hypotheses were investigated – H1: seeing a relevant debunking post will increase agreement with the target fact of the debunking, H2: higher trust in the source of the post will result in increased agreement with the target fact, and H3: A three-way interaction will exist such that when the myth being debunked is aligned with the specific expertise of the source, the impact of source trust will increase. Results show that, overall, the myth debunkings were ineffective, i.e., there were no significant differences in agreement with debunking target facts between any of the treatment conditions and the control condition; thus H1 was not supported. However, participants source trust increased belief in the target fact for the consensus and health message conditions, partially supporting H2 and indicating that source trust does play some kind of role in debunking effectiveness. Finally, H3 was not supported as no three-way interactions between source, myth, and source trust were significant. The lack of support for H1 is interesting, considering that many previous studies have shown debunking via social media to be effective. This null finding may be due to over-representation of Democrats in the sample, as well as a potential mediating role played by political party identification resulting in increased tendency of participants to think politically about climate change when their party affiliation is made salient. Results of this study indicate that further research is needed to explore the effect of source expertise and debunking topic alignment on debunking effectiveness. Despite the lack of support for H1, previous research and the partial support present for H2 indicate that trusted sources should continue climate myth debunking work