MD-SOAR Maryland Shared Open Access Repository
Not a member yet
34521 research outputs found
Sort by
P-Mix: A Data Augmentation Method for Contrastive Learning based Human Activity Recognition
No full textSupervised human activity recognition (HAR) with sensor data typically demands substantial labeled datasets to train robust models. Contrastive learning offers a self-supervised alternative by leveraging data augmentation to improve representation learning. However, most existing augmentation methods operate independently on either the time or channel dimension and often introduce unstructured noise, which can distort meaningful temporal and spectral patterns. To address these limitations, we present a novel P-Mix data augmentation method for contrastive learning in HAR tasks, specifically designed to be compatible with the SimCLR framework. P-Mix is a customized data augmentation method tailored to sensor data for human activity recognition, which slices and recombines both the time and channel dimensions, merging multiple temporal segments to encourage the model to explore the underlying relationships and variations in the data in an unsupervised setting. To capture motion cycles and long-term dependencies, we employ shorter temporal segments as fundamental processing units along the time dimension. By incorporating structured noise patterns based on motion cycle characteristics within these segments, we effectively enhance the model’s robustness and generalization capabilities. Extensive evaluations across five HAR benchmarks demonstrate that P-Mix achieves consistent improvements over the strongest baseline (Resample), delivering relative F1-score gains ranging from 1.87% (USC-HAD: 85.63% vs 83.93%) to 6.53% (DSADS: 97.24% vs 91.28%) through controlled multidimensional fusion. These results demonstrate the effectiveness of our approach in optimizing data generation and augmentation strategies for HAR tasks.This work was supported in part by the National Natural Science Foundation of China (62006110), the Natural Science Foundation of Hunan Province (2024JJ7428, 2023JJ30518) and the Scientific research project of Hunan Provincial Department of Education (22C0229).(Corresponding author:Tao Zhu.)https://ieeexplore.ieee.org/document/1113737
Connecting Language and Emotion in Large Language Models for Human-AI Collaboration
No full textLarge Language Models demonstrate linguistic abilities on par with humans, able to generate short texts, stories, instructions, and even code that’s often indistinguishable from what is created by humans. This allows humans to use large language models (LLMs) collaboratively — as communication aides or writing assistants. However, humans cannot always assume an LLM will behave the same way another person would. This is particularly evident in subjective scenarios such as where emotion is involved. In this work, I explore to what depth do LLMs perceive and understand human emotions, and look at ways of describing an emotion to an LLM for collaborative work. First, I study the problem of classifying emotions and show that LLMs perform well on their own, and can also improve smaller models at the same task. Secondly, I focus on generating emotions, using the problem space of keyword-constrained generation and a human participant-study to see where human expectations and LLM outputs diverge and how we can minimize any such misalignment. Here, I find that using English Words and Lexical expressions Valence-Arousal-Dominance (VAD) scales lead to good alignment and generation quality, while Numeric dimensions of VAD or Emojis fare worse
Discussion of the Article by Haslett, Isotalo, Markiewicz and Puntanen in Sankhya A: Revisiting some Results in C.R. Rao’s Paper in Sankhya in 1971
No full textThe article by Haslett et al. provides insights into the seminal contribution of Rao on the unified theory of linear estimation. This discussion provides additional commentary.https://link.springer.com/article/10.1007/s13171-025-00413-
GRB 221009A: the B.O.A.T Burst that Shines in Gamma Rays
No full textWe present a complete analysis of Fermi Large Area Telescope (LAT) data of GRB 221009A, the brightest gamma-ray burst (GRB) ever detected. The burst emission above 30 MeV detected by the LAT preceded, by 1 s, the low-energy (<10 MeV) pulse that triggered the Fermi Gamma-Ray Burst Monitor (GBM), as has been observed in other GRBs. The prompt phase of GRB 221009A lasted a few hundred seconds. It was so bright that we identify a bad time interval of 64 s caused by the extremely high flux of hard X-rays and soft gamma rays, during which the event reconstruction efficiency was poor and the dead time fraction quite high. The late-time emission decayed as a power law, but the extrapolation of the late-time emission during the first 450 s suggests that the afterglow started during the prompt emission. We also found that high-energy events observed by the LAT are incompatible with synchrotron origin, and, during the prompt emission, are more likely related to an extra component identified as synchrotron self-Compton (SSC). A remarkable 400 GeV photon, detected by the LAT 33 ks after the GBM trigger and directionally consistent with the location of GRB 221009A, is hard to explain as a product of SSC or TeV electromagnetic cascades, and the process responsible for its origin is uncertain. Because of its proximity and energetic nature, GRB 221009A is an extremely rare event.The Fermi-LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat `a l’Energie Atomique and the Centre National de la Recherche Scientifique / Institut National de Physique Nucl´eaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare inItaly, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), High Energy Accelerator Research Organization (KEK) and Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden. Additional support for science analysis during the operations phase is gratefully acknowledged from the Istituto Nazionale di Astrofisica in Italy and the Centre National d’Etudes Spatiales in France. This work performed in part under DOE Contract DE-AC02-76SF00515.https://iopscience.iop.org/article/10.3847/1538-4365/ada272/met
Improving Gamma Imaging in Proton Therapy by Sanitizing Compton Camera Simulated Patient Data using Neural Networks through the BRIDE Pipeline
No full text2024 IEEE International Conference on Big Data (BigData), 15-18 December 2024, Washington, DC, USAPrecision medicine in cancer treatment increasingly relies on advanced radiotherapies, such as proton beam radiotherapy, to enhance efficacy of the treatment. When the proton beam in this treatment interacts with patient matter, the excited nuclei may emit prompt gamma ray interactions that can be captured by a Compton camera. The image reconstruction from this captured data faces the issue of mischaracterizing the sequences of incoming scattering events, leading to excessive background noise. To address this problem, several machine learning models such as Feedfoward Neural Networks (FNN) and Recurrent Neural Networks (RNN) were developed in PyTorch to properly characterize the scattering sequences on simulated datasets, including newly-created patient medium data, which were generated by using a pipeline comprised of the GEANT4 and Monte-Carlo Detector Effects (MCDE) softwares. These models were implemented using the novel ‘Big-data REU Integrated Development and Experimentation’ (BRIDE) platform, a modular pipeline that streamlines preprocessing, feature engineering, and model development and evaluation on parallelized GPU processors. Hyperparameter studies were done on the novel patient data as well as on water phantom datasets used during previous research. Patient data was more difficult than water phantom data to classify for both FNN and RNN models. FNN models had higher accuracy on patient medium data but lower accuracy on water phantom data when compared to RNN models. Previous results on several different datasets were reproduced on BRIDE and multiple new models achieved greater performance than in previous research.This work is supported by the grant “REU Site: Online Interdisciplinary Big Data Analytics in Science and Engineering” from the National Science Foundation (grant no. OAC– 2348755). Undergraduate assistant co-author Obe acknowledges support from an REU Supplement. Co-authors Sharma and Ren acknowledge support from the NIH. The hardware used in the computational studies is part of the UMBC High Performance Computing Facility (HPCF). The facility is supported by the U.S. National Science Foundation through the MRI program (grant nos. CNS–0821258, CNS–1228778, OAC–1726023, and CNS–1920079) and the SCREMS program (grant no. DMS–0821311), with additional substantial support from the University of Maryland, Baltimore County (UMBC). See hpcf.umbc.edu for more information on HPCF and the projects using its resources.https://ieeexplore.ieee.org/document/10825318
GPS-Health: A Novel Analytic Infrastructure for Capturing, Visualizing, and Analyzing Multi-Level, Multi-Domain Geographically Distributed Social Determinants of Health
No full textBackground Health disparities across a range of conditions and outcomes exist across the life course and are driven by the uneven geographic distribution of multidimensional social determinants of health (SDOH). Previous multidimensional measures of SDOH (e.g. Area Deprivation Index, Social Vulnerability Index, Social Deprivation Index) collapse multiple measures into a single summary value applied to everyone living within a predefined map unit, engendering construct and internal validity issues.Methods We present a new SDOH data approach: the Geographic Patterns of Social Determinants of Health (GPS-Health). We use a theoretical framework weaving together kyriarchy, intersectionality, and structural violence to select SDOH domains that can elucidate how individuals experience multidimensional spatial distributions of SDOH. We apply the approach to Maryland.Results Our dataset includes 2,369,365 property parcels, from which we calculate distances to 8 types of SDOH exact locations.Discussion GPS-Health will aid in the understanding of how the SDOH influence individual health outcomes.SJH is funded on a NIDDK institutional training grant: 5T32DK098107-09. All authors are investigators at the University of Maryland-Institute for Health Computing, which is supported by funding from Montgomery County, Maryland and The University of Maryland Strategic Partnership: MPowering the State, a formal collaboration between the University of Maryland, College Park and the University of Maryland, Baltimore. EMD is a member of the United States Preventive Services Task Force (USPSTF), this article does not necessarily represent the views and policies of the USPSTF. Research reported in this publication was supported by the National on Minority Health and Health Disparities (R01MD015716 [TTN]): the content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.https://www.medrxiv.org/content/10.1101/2025.01.03.25319962v
Describing seasonal mixtures of cloud regimes via “regimes of regimes”
No full textWe propose a new type of cloud classification, relevant to monthly or longer time scales, but which inherently still encompasses cloud subgrid variability information at ~100 km scales. Our proposed classification partitions frequencies of occurrence over these scales of previously defined cloud regimes (CRs). We call the resulting distinct cloud entities regimes of regimes (RORs). While the CRs have been previously shown to successfully classify daily mesoscale subgrid variability via distributions of cloud fraction within distinct combinations of cloud top pressure and cloud optical thickness, the RORs essentially represent the prevalent seasonal mixtures of these CRs. RORs thus embody the seasonal cloudiness of a mesoscale region. We show that each ROR can still be associated with more traditional cloud classifications via composites of coincident active (lidar and cloud radar) cloud views. In a first application that gauges the potential utility of RORs, we pair them with CERES EBAF radiative fluxes to gain insight into recent trends of the cloud radiative effect. The ROR corresponding to an environment of shallow convection stands out in this analysis largely because of its declining population. Our study demonstrates the potential of RORs to categorize globally mesoscale cloudiness at monthly/seasonal scales and to serve as proxies of different atmospheric states at these scales.Resources supporting this work were provided by the NASA High-End Computing (HEC) Program through the NASA Center for Climate Simulation (NCCS) at Goddard Space Flight Center. Support by NASA’s CloudSat-CALIPSO Science Team and MEaSUREs programs is gratefully acknowledged.https://journals.ametsoc.org/view/journals/clim/aop/JCLI-D-24-0275.1/JCLI-D-24-0275.1.xm
Temperature-Dependent Properties of Atomic Layer Deposition-Grown TiO2₂Thin Films
No full textThis study investigates the presence of titanium oxynitride bonds in titanium dioxide (TiO₂) thin films grown by atomic layer deposition (ALD) using tetrakis dimethyl amino titanium (TDMAT) and water at temperatures between 150 and 350 °C and its effect on the films’ optical and electrical properties. Compositional analysis using X-ray photoelectron spectroscopy (XPS) reveals increased incorporation of oxynitride bonds as the process temperature increases. Furthermore, depth profile data demonstrates an increase in the abundance of this type of bonding from the surface to the bulk of the films. Ultraviolet-visible spectroscopy (UV-vis) measurements correlate increased visible light absorption for the films with elevated oxynitride incorporation. The optical constants (n, k) of the films show a pronounced dependence on the process temperature that is mirrored in the film conductivity. The detection of oxynitride bonding suggests a secondary reaction pathway in this well-established ALD process chemistry, that may impact film properties. These findings indicate that the choice of process chemistry and conditions can be used to optimize film properties for optoelectronic applications.The author would like to thank Dr. Ilkeun Lee from the University of California, Riverside (UCR) for the XPS measurements and Dr. Tanguy Terlier from Rice University for the ToF-SIMS measurements. The author thanks Andrea Donohue from Woollam for her assistance with the optical models. The Authors were grateful to Dr. Lisa Kelly and her students from the Department of Chemistry and Biochemistry at UMBC for their assistance with obtaining the transmission data. The Author kindly acknowledges Dr. Karen Gaskell and Dr. Peter Zavalij from the University of Maryland, College Park (UMD) for their help in obtaining AFM and XRD data. The author also acknowledges the support of the Maryland NanoCenter and its FabLab. Student support was provided by an NIGMS Graduate Research Training Initiative for Student Enhancement (G-RISE) Grant (T32- GM144876). This material was based upon work supported by the National Science Foundation (NSF) under a grant from the Directorate for Mathematical and Physical Sciences (DMR)-1905305.https://onlinelibrary.wiley.com/doi/abs/10.1002/admi.20240085
Performance of the 0-padding Optimal Filter Method in Non-linear Gain Calibration
No full textThe focal-plane detector, the X-ray Integral Field Unit (X-IFU), on-board ESA's Athena space telescope is a transition edge sensor (TES) microcalorimeter array with 1.5k pixels, designed to provide spatially-resolved, high-resolution spectroscopy over the energy range 0.2-12 keV. The onboard event processor uses a digital optimal filter to determine the pulse-height of the measured current pulse from every X-ray photon striking the array. A modified optimal filter called the 0-padding filter has recently been proposed. This is a truncated version of the standard optimal filter and has been shown to provide comparable energy resolution but with the benefit of reduced computational expense. Whereas the standard optimal filter has zero integral and is not sensitive to variations in the DC level of the measured signal, the integral of the 0-padded version is non-zero and thus is more sensitive to fluctuations in DC signal over time. In this work, we explore the effect of 0-padding on the energy scale calibration using data from 250-pixels in a prototype Athena X-IFU array, measured over the range 1.3-12 keV.This material is based upon work supported by NASA under award number 80GSFC21M0002. N. Cardiel, M. T. Ceballos, and B. Cobo acknowledge Grant PID2021 122955OB-C41 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”.https://ieeexplore.ieee.org/abstract/document/1084967
Transformer Models in Education: Summarizing Science Textbooks with AraBART, MT5, AraT5, and mBART
No full textRecently, with the rapid development in the fields of technology and the increasing amount of text t available on the internet, it has become urgent to develop effective tools for processing and understanding texts in a way that summaries the content without losing the fundamental essence of the information. Given this challenge, we have developed an advanced text summarization system targeting Arabic textbooks. Relying on modern natu-ral language processing models such as MT5, AraBART, AraT5, and mBART50, this system evaluates and extracts the most important sentences found in biology textbooks for the 11th and 12th grades in the Palestinian curriculum, which enables students and teachers to obtain accurate and useful summaries that help them easily understand the content. We utilized the Rouge metric to evaluate the performance of the trained models. Moreover, experts in education Edu textbook authoring assess the output of the trained models. This approach aims to identify the best solutions and clarify areas needing improvement. This research provides a solution for summarizing Arabic text. It enriches the field by offering results that can open new horizons for research and development in the technologies for understanding and generating the Arabic language. Additionally, it contributes to the field with Arabic texts through creating and compiling schoolbook texts and building a dataset.https://arxiv.org/abs/2406.0769