UMBC Student Collection

Permanent URI for this collection


Recent Submissions

Now showing 1 - 20 of 3243
  • Item
    KiNETGAN: Enabling Distributed Network Intrusion Detection through Knowledge-Infused Synthetic Data Generation
    (2024-05-26) Kotal, Anantaa; Luton, Brandon; Joshi, Anupam
    In the realm of IoT/CPS systems connected over mobile networks, traditional intrusion detection methods analyze network traffic across multiple devices using anomaly detection techniques to flag potential security threats. However, these methods face significant privacy challenges, particularly with deep packet inspection and network communication analysis. This type of monitoring is highly intrusive, as it involves examining the content of data packets, which can include personal and sensitive information. Such data scrutiny is often governed by stringent laws and regulations, especially in environments like smart homes where data privacy is paramount. Synthetic data offers a promising solution by mimicking real network behavior without revealing sensitive details. Generative models such as Generative Adversarial Networks (GANs) can produce synthetic data, but they often struggle to generate realistic data in specialized domains like network activity. This limitation stems from insufficient training data, which impedes the model’s ability to grasp the domain’s rules and constraints adequately. Moreover, the scarcity of training data exacerbates the problem of class imbalance in intrusion detection methods. To address these challenges, we propose a Privacy-Driven framework that utilizes a knowledge-infused Generative Adversarial Network for generating synthetic network activity data (KiNETGAN). This approach enhances the resilience of distributed intrusion detection while addressing privacy concerns. Our Knowledge Guided GAN produces realistic representations of network activity, validated through rigorous experimentation. We demonstrate that KiNETGAN maintains minimal accuracy loss in downstream tasks, effectively balancing data privacy and utility.
  • Item
    Determination of Planetary Boundary Layer Heights on Short Spatial and Temporal Scales from Surface and Airborne Vertical Profilers during DISCOVER-AQ
    (2022-09) Delgado, Ruben; Berkoff, T; Compton, J.; St Pe, Alexandra E.; Baker, Barry; Hoff, Raymond M.; Martins, Douglas K.; Thompson, Anne M.; Yang, Su; Christopher, Sundar A.; Joseph, Everette; Tzortziou, Maria; Landry, Laura; Woodman, Michael; Lolli, Simone; Weinheimer, Andrew J.; Montzka, Denise D.; Knapp, David J.; Ferrare, Richard A.; Hostetler, Chris A.; Crawford, James
  • Item
    Biclustering a dataset using photonic quantum computing
    (2024-05-28) Borle, Ajinkya; Bhave, Ameya
    Biclustering is a problem in machine learning and data mining that seeks to group together rows and columns of a dataset according to certain criteria. In this work, we highlight the natural relation that quantum computing models like boson and Gaussian boson sampling (GBS) have to this problem. We first explore the use of boson sampling to identify biclusters based on matrix permanents. We then propose a heuristic that finds clusters in a dataset using Gaussian boson sampling by (i) converting the dataset into a bipartite graph and then (ii) running GBS to find the densest sub-graph(s) within the larger bipartite graph. Our simulations for the above proposed heuristics show promising results for future exploration in this area.
  • Item
    Bay Breeze Impact on Surface Ozone at Edgewood, Maryland
    (2022-09) Stauffer, R. M.; Thompson, Anne M.; Martins, D. K.; Clark, R.; Herman, Jay; Berkoff, T.; Baker, Barry; Delgado, Ruben
  • Item
    Scoping: Towards Streamlined Entity Collections for Multi-Sourced Entity Resolution with Self-Supervised Agents
    (SciTePress, 2024) Traeger, Leonard; Behrend, Andreas; Karabatis, George
    Linking multiple entities to a real-world object is a time-consuming and error-prone task. Entity Resolution (ER) includes techniques for vectorizing entities (signature), grouping similar entities into partitions (blocking), and matching entity pairs based on specified similarity thresholds (filtering). This paper introduces scoping as a new and integral phase in multi-sourced ER with potentially increased heterogeneity and more unlinkable entities. Scoping reduces the space of candidate entity pairs by ranking, detecting, and removing unlinkable entities through outlier algorithms and reusable self-supervised autoencoders, leaving intact the set of true linkages. Evaluations on multi-sourced schemas show that autoencoders perform best in schemas relevant to each other, where they reduce entity collections to 77% and still contain all linkages.
  • Item
    Evaluating Machine Learning and Statistical Models for Greenland Bed Topography
    Yi, Katherine; Dewar, Angelina; Tabassum, Tartela; Lu, Jason; Chen, Ray; Alam, Homayra; Faruque, Omar; Li, Sikan; Morlighem, Mathieu; Wang, Jianwu
    The purpose of this research is to study how different machine learning and statistical models can be used to predict bed topography in Greenland using ice-penetrating radar and satellite imagery data. Accurate bed topography representations are crucial for understanding ice sheet stability, melt, and vulnerability to climate change. We explored nine predictive models including dense neural network, LSTM, variational auto-encoder (VAE), extreme gradient boosting (XGBoost), gaussian process regression, and kriging based residual learning. Model performance was evaluated with mean absolute error (MAE), root mean squared error (RMSE), coefficient of determination (R2), and terrain ruggedness index (TRI). In addition to testing various predictive models, different interpolation methods, including Nearest Neighbor interpolation, Bilinear Interpolation, and Universal Kriging were used to obtain estimates the values of ice surface features at the ice bed observation locations. The XGBoost model with Universal Kriging interpolation exhibited strong predictive capabilities but demands extensive resources. Alternatively, the XGBoost model with bilinear interpolation showed robust predictive capabilities and required fewer resources. These models effectively captured the complexity of the Greenland ice sheet terrain with precision and efficiency, making them valuable tools for representing spatial patterns in diverse landscapes.
  • Item
    Predicting Ice-bed Topography using Predictive Modeling
    (2024-05-14) Alam, Homayra; Yi, Katherine; Dewar, Angelina; Tabassum, Tartela; Lu, Jason; Chen, Ray; Faruque, Omar; Li, Sikan; Morlighem, Mathieu
    The purpose of this research is to study how different machine learning and statistical models can be used to predict bedrock topography under the Greenland ice sheet using ice-penetrating radar and satellite imagery data. Accurate bed topography representations are crucial for understanding ice sheet stability and vulnerability to climate change. We explore nine predictive models including dense neural network, long-short term memory, variational auto-encoder, extreme gradient boosting (XGBoost), gaussian process regression, and kriging based residual learning. Model performance is evaluated with mean absolute error (MAE), root mean squared error (RMSE), coefficient of determination (R ² ), and terrain ruggedness index (TRI). In addition to testing various models, different interpolation methods, including nearest neighbor, bilinear, and kriging, are also applied in preprocessing. The XGBoost model with kriging interpolation exhibit strong predictive capabilities but demands extensive resources. Alternatively, the XGBoost model with bilinear interpolation shows robust predictive capabilities and requires fewer resources. These models effectively capture the complexity of the terrain hidden under the Greenland ice sheet with precision and efficiency, making them valuable tools for representing spatial patterns in diverse landscapes.
  • Item
    Fluorophore-Induced Plasmonic Current Generation from Copper Nanoparticle Films
    (ACS, 2024-05-24) Pierce, Daniel; Saha, Lahari; Geddes, Chris
    We describe the process of generating a fluorophore-induced plasmonic current (FIPC) from copper nanoparticle films. Previous work and the literature have shown that excited near-field fluorophores are able to plasmonically couple with metal nanoparticle films (MNFs), inducing surface plasmons in the films. These induced surface plasmons are then in turn able to generate a directly measurable electrical current across the film. These generated currents have been quantified and detected in noble metal films, such as those made from Ag and Au, but due to the cost of such films, there has been a push to use lower cost materials for FIPC. Previous work has detailed the use of gold, silver, and aluminum films for these purposes, and in this paper, we will subsequently examine the ability of thermally deposited copper films to generate FIPC when in close proximity to excited near-field fluorophores. We report the effects of copper film thickness, the effects of light polarization and solution conductance, and the effects of metal-enhanced fluorescence (MEF) emission on the generation of plasmonic current.
  • Item
    MPL Planetary Boundary Layer Heights at Golden, Fort Collins and Platteville
    (2015-05-05) Hoff, Raymond; Berkoff, T.; Sullivan, J.; Orozco, Daniel; Delgado, Ruben; Clarke, R.; Thompson, Anne M.
    Three micropulse lidars and one elastic lidar were employed at three sites in order to get backscatter, extinction, and PBL height estimates. We show here the PBL comparisons between the lidars and the aircraft profiles. PBL Heights are retrieved from aerosol gradients typically which occur at the top of the PBL where it interacts with a clearer free troposphere. In Colorado, this was not always the case as air aloft had substantially the same aerosol load on many days. The PBL retrieval uses the Haar wavelet to determine this gradient and the parameters for this wavelet were adjusted for aerosol loading, heating and time of day.
  • Item
    Strong Scalability Studies for the 2-D Poisson Equation on the Taki 2021 Cluster with Historical Comparison
    (2024) Shakeri, Ehsan; Gobbert, Matthias
    The new 2021 nodes in the cluster taki in the UMBC High Performance Computing Facility contain two 24core Intel Cascade Lake CPUs and 192 GB of memory per node, connected by an high-performance InfiniBand interconnect. Parallel performance studies for the memory-bound test problem of the Poisson equation in two spatial dimensions yield several conclusions: Strong scalability studies demonstrate excellent performance when using multiple nodes due to the low latency of the high-performance interconnect and good speedup when using all cores of the multi-core CPUs. For the largest numbers of processes per node, the runtime for the code is not significantly reduced, a typical behavior characteristic of memory-bound code. Comparisons to results on past clusters in HPCF bring out that core-per-core performance of serial code improvements has improved again, demonstrating the quality of the newest CPUs. Also node-per-node performance of parallel code continues to improve due to the larger number of cores available on a node. albeit we have fewer nodes in taki 2021 than we had in previous partitions.
  • Item
    “This app said I had severe depression, and now I don’t know what to do”: the unintentional harms of mental health applications
    (ACM, 2024-05-11) Kang, Rachael; Reynolds, Tera L
    A growing market for mental health applications and increasing evidence for the efficacy of these applications have made apps a popular mode of mental healthcare delivery. However, given the gravity of mental illnesses, the potential harms of using these applications must be continually investigated. In this study, we conducted a thematic analysis using user-comments left on depression self-management applications. We analyzed 6,253 reviews from thirty-six, systematically selected apps from the Google Play and Apple App stores. We identified four themes regarding the potential, unintentional harms caused by these applications. This study uniquely contributes to the literature by examining the reported harms to users caused by depression self-management apps and contextualizing them in an ethical framework. We provide recommendations to developers for creating ethical depression self-management apps and resources for practitioners and consumers to aid in screening apps.
  • Item
    "That's Kind of Sus(picious)": The Comprehensiveness of Mental Health Application Users' Privacy and Security Concerns
    (ACM, 2024-05-11) Khoo, Yi Xuan; Kang, Rachael; Reynolds, Tera L; Mentis, Helena
    With the increasing usage of mental health applications (MHAs), there is growing concern regarding their data privacy practices. Analyzing 437 user reviews from 83 apps, we outline users' predominant privacy and security concerns with currently available apps. We then compare those concerns to criteria from two prominent app evaluation websites – Privacy Not Included and One Mind PsyberGuide. Our findings show that MHA users have myriad data privacy and security concerns including a user's control over their own data, but these concerns do not often overlap with those of experts from evaluation websites who focus more on issues such as required password strength. We highlight this disconnect and propose solutions in how the mental health care ecosystem can provide better guidance to MHA users and experts from the fields of privacy and security and mental health technology in choosing and evaluating, respectively, potentially useful mental health apps.
  • Item
    Motives and Role of Psychological Ownership in AR Workspaces for Remote Collaboration
    (ACM, 2024-05-11) Seo, Jwawon
    Augmented Reality (AR) systems have been proposed as solutions to reinforce remote collaboration over performing physical tasks by integrating digital information into the physical environment. However, there is a fundamental gap that arises from the difference in the degree to which remote and local workers interact with virtual and physical objects, which is attributed to differing senses of ownership over these objects. Addressing this gap, my thesis investigates how collaborative functionality affects psychological ownership in shared AR workspaces. Specifically, the study examines how the ability to modify annotations shapes Individual and Collective Psychological Ownership (IPO and CPO). Furthermore, I explore how IPO and CPO mediate the effect of collaborative functionality on the outcomes of remote collaboration. My overall goal is to extend the theoretical understanding of psychological ownership, offering insights into the dynamics of shared AR environments by focusing on the interplay between physical and virtual elements.
  • Item
    PeerConnect: Co-Designing a Peer-Mentoring Support System with Computing Transfer Students
    (ACM, 2024-05-11) Anthraper, Nisha; Javiya, Prachee; Iluru, Sai; Chen, Lujie Karen; Kleinsmith, Andrea
    In the US, nearly half of the STEM undergraduates begin their academic careers at community colleges. Transferring to four-year institutions can be challenging. Evidence suggests that mentoring can help by increasing a sense of belonging and retention. We engaged mentors and mentees from a pilot mentoring program for new transfer students in computing majors at a minority-serving institution in the Northeastern US in a co-design workshop to understand their needs and requirements for a peer-mentoring system, PeerConnect. PeerConnect aims to foster transfer students's academic and social engagement, increase self-efficacy and belonging, and develop students' self-regulated learning skills. Preliminary results show that students want features that push the system beyond merely measuring engagement to actively promoting it. This study contributes to HCI and CSCW work in designing support systems for mentoring and peer support programs in educational settings and to the emerging literature on student-centered learning analytics systems.
  • Item
    Deep PIM kinase substrate profiling reveals new rational co-therapeutic strategies for acute myeloid leukemia
    (American Society of Hematology, 2024-05-13) Joglekar, Tejashree; Chin, Alexander; Voskanian-Kordi, Alin; Seungchul, Baek; Raja, Azim; Rege, Apurv; Huang, Weiliang; Kane, Maureen A; Laiho, Marikki; Webb, Thomas; Fan, Xiaoxuan; Rubenstein, Michael; Bieberich, Charles J.; Li, Xiang
    Provirus integration site for Moloney murine leukemia virus (PIM) family serine/threonine kinases perform pro-tumorigenic functions in hematologic malignancies and solid tumors by phosphorylating substrates involved in tumor metabolism, cell survival, metastasis, inflammation, and immune cell invasion. However, a comprehensive understanding of PIM kinase functions is currently lacking. Multiple small molecule PIM kinase inhibitors are currently being evaluated as co-therapeutics in cancer patients. To further illuminate PIM kinase functions in cancer, we deeply profiled PIM1 substrates using the reverse in-gel kinase assay to identify downstream cellular processes targetable with small molecules. Pathway analyses of putative PIM substrates nominated RNA splicing and rRNA processing as PIM-regulated cellular processes. PIM inhibition elicited reproducible splicing changes in PIM-inhibitor-responsive acute myeloid leukemia (AML) cell lines. PIM inhibitors synergized with splicing modulators targeting splicing factor 3b subunit 1 (SF3B1) and serine-arginine protein kinase 1 (SRPK1) to kill AML cells. PIM inhibition also altered rRNA processing, and PIM inhibitors synergized with an RNA polymerase I inhibitor to kill AML cells and block AML tumor growth. These data demonstrate that deep kinase substrate knowledge can illuminate unappreciated kinase functions, nominating synergistic co-therapeutic strategies. This approach may expand the co-therapeutic armamentarium to overcome kinase-inhibitor resistant disease that limits durable responses in malignant disease.
  • Item
    v-Palindromes: An Analogy to the Palindromes
    (2024-04-24) Bispels, Chris; Boran, Muhammet; Miller, Steven J.; Sosis, Eliel; Tsai, Daniel
    Around the year 2007, one of the authors, Tsai, accidentally discovered a property of the number 198 he saw on the license plate of a car. Namely, if we take 198 and its reversal 891, which have prime factorizations 198 = 2 · 3² · 11 and 891 = 3⁴ · 11 respectively, and sum the numbers appearing in each factorization getting 2 + 3 + 2 + 11 = 18 and 3 + 4 + 11 = 18, both sums are 18. Such numbers were later named v-palindromes because they can be viewed as an analogy to the usual palindromes. In this article, we introduce the concept of a v-palindrome in base b and prove their existence for infinitely many bases. We also exhibit infinite families of v-palindromes in bases p + 1 and p² + 1, for each odd prime p. Finally, we collect some conjectures and problems involving v-palindromes.
  • Item
    The people behind the papers – Jason Ko and Daniel Lobo
    (The Company of Biologists Ltd, 2024-05-09) Ko, Jason; Lobo, Daniel
    Planarians grow when they are fed and shrink during periods of starvation. However, it is unclear how they maintain appropriate body proportions as their size changes. A new paper in Development investigates the differences between growth and shrinkage dynamics and builds a mathematical model to explore the mechanisms underpinning these two processes. To learn more about the story behind the paper, we caught up with first author, Jason Ko, and corresponding author, Daniel Lobo, Associate Professor at the University of Maryland.
  • Item
    Generalized constrained energy minimization approach to subpixel target detection for multispectral imagery
    (SPIE, 2000-05-01) Chang, Chein-I; Liu, JihMing; Chieu, BinChang; Ren, Hsuan; Wang, Chuin-Mu; Lo, ChienShun; Chung, Pau-Choo; Yang, Ching-Wen; Ma, DyeJyun
    Subpixel detection in multispectral imagery presents a challenging problem due to relatively low spatial and spectral resolution. We present a generalized constrained energy minimization (GCEM) approach to detecting targets in multispectral imagery at subpixel level. GCEM is a hybrid technique that combines a constrained energy minimization (CEM) method developed for hyperspectral image classification with a dimensionality expansion (DE) approach resulting from a generalized orthogonal subspace projection (GOSP) developed for multispectral image classification. DE enables us to generate additional bands from original multispectral images nonlinearly so that CEM can be used for subpixel detection to extract targets embedded in multispectral images. CEM has been successfully applied to hyperspectral target detection and image classification. Its applicability to multispectral imagery is yet to be investigated. A potential limitation of CEM on multispectral imagery is the effectiveness of interference elimination due to the lack of sufficient dimensionality. DE is introduced to mitigate this problem by expanding the original data dimensionality. Experiments show that the proposed GCEM detects targets more effectively than GOSP and CEM without dimensionality expansion.
  • Item
    Unsupervised hyperspectral image analysis with projection pursuit
    (IEEE, 2000-11) Ifarraguerri, A.; Chang, Chein-I
    Principal components analysis (PCA) is effective at compressing information in multivariate data sets by computing orthogonal projections that maximize the amount of data variance. Unfortunately, information content in hyperspectral images does not always coincide with such projections. The authors propose an application of projection pursuit (PP), which seeks to find a set of projections that are "interesting," in the sense that they deviate from the Gaussian distribution assumption. Once these projections are obtained, they can be used for image compression, segmentation, or enhancement for visual analysis. To find these projections, a two-step iterative process is followed where they first search for a projection that maximizes a projection index based on the information divergence of the projection's estimated probability distribution from the Gaussian distribution and then reduce the rank by projecting the data onto the subspace orthogonal to the previous projections. To calculate each projection, they use a simplified approach to maximizing the projection index, which does not require an optimization algorithm. It searches for a solution by obtaining a set of candidate projections from the data and choosing the one with the highest projection index. The effectiveness of this method is demonstrated through simulated examples as well as data from the hyperspectral digital imagery collection experiment (HYDICE) and the spatially enhanced broadband array spectrograph system (SEBASS).
  • Item
    Target-constrained interference-minimized approach to subpixel target detection for hyperspectral images
    (SPIE, 2000-12-01) Ren, Hsuan; Chang, Chein-I
    Due to significantly improved spatial and spectral resolution, hyperspectral sensors can now detect many substances that cannot be resolved by multispectral sensors. However, this comes at the price that many unknown and unidentified signal sources, referred to as interferers, may also be extracted unexpectedly. Such interferers generally produce additional noise effects on target detection and must therefore be taken into account. The problem associated with this interference is challenging because its nature is generally unknown and it cannot be identified from an image scene. This paper presents an approach, called the target-constrained interference-minimized filter (TCIMF), which does not require one to identify interferers, but can minimize the effects caused by interference. It designs a finite-impulse-response filter that specifies targets of interest in such a way that the desired targets and undesired targets will be passed through and rejected by the filter, respectively; the filter output energy resulting from unknown signal sources is also minimized. More precisely, the TCIMF accomplishes three tasks simultaneously: detection of the desired targets, elimination of the undesired targets, and minimization of interfering effects. A recently developed technique, constrained energy minimization (CEM), can be considered as a suboptimal version of the TCIMF. Computer simulations and hyperspectral image experiments are conducted to demonstrate advantages of the TCIMF over the CEM.