26419 research outputs found
Sort by
Computational Modeling for Automatic Superconducting Cavity Fault Prediction and Classification Using Time Series Signals
No full textProcessing multivariate time series signals collected from sensor networks is challenging because of complex temporal dependencies and non-stationarity. With the advent of artificial intelligence (AI) like machine learning and deep learning, it has become possible to process sensor-driven time series data more effectively than traditional statistical methods.
This dissertation aims to develop machine learning and deep learning models to address machine fault diagnosis using multivariate time series signals collected from the Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Lab. The first goal of the proposed work is to develop deep learning–based classification models and an unsupervised fault clustering approach using multivariate time series data. For the classification task, we designed a hybrid architecture that combines Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNN) to classify superconducting radio frequency (SRF) faults and cavity types. For the clustering task, we proposed a multivariate time series clustering approach based on Shapelet learning to identify fault patterns without requiring labeled data. The proposed model outperforms two other unsupervised clustering approaches, offering superior performance. The second goal of the proposed work aims to develop an SRF cavity fault prediction model using multivariate time series signals. In this work, we develop an uncertainty-aware hybrid deep learning model that combines LSTM and CNN architecture to predict SRF cavity fault using pre-fault signals. We incorporate multiple consecutive windows fault prediction and adjusting the fault confidence threshold strategies to enhance the fault prediction performance in real-time implementation. The model is evaluated using a highly imbalanced dataset from CEBAF, which reflects the real-world operating conditions. The results of the proposed approach may offer the potential to take preventive action before a failure occurs. The third goal of the proposed work aims to develop a domain adaptation network that performs fault predictions with changing data scenarios. We propose a Siamese network using a hybrid LSTM-CNN architecture to perform this task. The model performance is compared with another transfer learning model and a model with no transfer learning. The Siamese network-based domain adaptation approaches perform better than other approaches under changing machine dynamics
Analysis of Multi Grade Deep Learning
No full textMulti-Grade Deep Learning (MGDL) is a training framework that incrementally builds deep neural networks. It does this by dividing the training process into multiple “grades,” where each grade sequentially trains a shallow neural network to learn the residue from the previous one, using the outputs of prior grades as input. This approach progresses from shallow to deep architectures. This dissertation offers a comprehensive theoretical and numerical analysis of the MGDL methodology.
We first demonstrate that MGDL can effectively learn target functions within the sum-composition learning format. In this context, MGDL approximates high-frequency components by composing multiple low-frequency functions. This unique capability explains how MGDL addresses the spectral bias commonly observed in standard deep neural networks, which typically struggle to learn high-frequency features. Extensive numerical experiments confirm that MGDL significantly improves the learning of these high-frequency components in target functions.
Next, we analyze the gradient descent method for training deep neural networks and highlight MGDL’s advantages over standard architectures. Specifically, we show that when each “grade” in MGDL consists of a single hidden layer with a piecewise linear activation function, it transforms a highly nonconvex optimization problem into a sequence of convex subproblems. This transformation provides a theoretical explanation for MGDL’s improved trainability.
We then apply MGDL to image reconstruction tasks, proposing a proximal gradient method to solve the resulting optimization problem and providing a convergence analysis of the algorithm. Experimental results indicate that MGDL achieves more stable training compared to standard deep neural networks. To understand this enhanced stability, we conduct an eigenvalue analysis of the iterative matrices arising from gradient descent. This analysis reveals that training stability is governed by the distribution of these eigenvalues. Numerical experiments show that standard deep neural networks often produce eigenvalues outside the convergence region, while MGDL consistently maintains eigenvalues within it, thereby explaining its superior stabilit
Legitimacy, War, and the State: A Framework for Understanding Legitimation Strategies in US Foreign Policy
No full textDo social norms play a role in policy maker’s foreign policy choices? What methods or strategies are effective in generating sustainable public support? In contrast to certain theories that conceive of international relations as purely divorced from social norms – be they international or domestic - I argue that the characteristic of legitimacy is critical to understanding how states construct and pursue their foreign policy. Using a methodology of historical review I examine four primary qualities of legitimation theory: Concern, Audience, Effort, and Reception; using the test case of the Americanization of the Vietnam War (1961-1965) this study will compare and contrast two Structuralist accounts of legitimation strategies: World Systems Theory (WST) and Structural Realism (SR). I argue that Structural Realism is inadequate in its grasp of legitimation within American foreign policy history, able to make some solid predictions on the international dimension but crippled in virtue of its de-emphasis of domestic politics. In contrast, this study argues that WST is able to incorporate a more comprehensive understanding of legitimation due to its inclusion of domestic politics as well as its modeling of ideational and social norms. However like SR it remains crippled due to its lack of First Image analysis which I argue plays a critical historical role in explaining the Americanization of the Vietnam War, even if both theories lack the theoretical capacity for it. This study ultimately strengthens the broader hypothesis reflected in Legitimation Theory that domestic and international audiences are critical in how countries construct and sustain their foreign policy, particularly through the use (or lack of) ideational norms that can transcend the domestic/international divide. Beyond analysis, I recommend that policy makers proactively seek to understand the role of domestic legitimacy and the need to effectively legitimize domestic audiences on behalf of a given foreign policy goal
A Focused Ethnography: Mental Health Therapists in Pediatric Integrated Care
No full textThe scope of practice for Licensed Professional Counselors (LPCs) and counseling residents (LPC-R) in integrated behavioral health (IBH) settings is grounded in accredited, theory-informed education and training. The benefits of IBH models are well-documented, however there is limited research that addresses how counselors develop professionally and clinically function, while navigating within pediatric inpatient environments. This qualitative dissertation used a focused ethnographic design to explore the lived experiences of ten purposively sampled counselors (n = 10) through demographic surveys, field observation, and semi-structured interviews. Reflexive journaling bracketed researcher bias, and thematic analysis, supported by NVivo and new member checking, guided interpretation.
Participants described navigating systemic and relational complexities, including fragmented supervision, shifting leadership, and marginalization within medical hierarchies. Simultaneously, they found meaning through peer consultation, collaborative teams, and witnessing patient and family system progress. Eight themes emerged: notable events, professional identity, conflict dynamics, collaboration, team cohesion, power imbalance, cultural relations, and continuous improvement.
Findings highlight how cultural dynamics, interdisciplinary relationships, and counselor voice shape clinical identity, therapeutic practice and ethical decision-making. This study emphasizes the need for experiential training, context-specific supervision, and mentorship within high-acuity counseling. Implications extend to counselor education, quality clinical supervision, and interdisciplinary workforce development to support the advancement of integrated care for vulnerable child and adolescent populations
Understanding and Helping International Students Cope When Catastrophes Strike in Their Home Countries: A Qualitative Study
No full textThis study explores the experiences of international students in the United States who come from countries that have been, or are being, affected by catastrophic events. The aim of the study is to understand the challenges such students face and to formulate supportive strategies that address these challenges. Utilizing the Interpretative Phenomenological Analysis (IPA), the study delves into the lived experiences of five international PhD students from Turkey, Ukraine, and Iran—nations that have recently experienced significant disasters or unrest. The study highlights the need for the development of effective personalized support from academic institutions, faculty, and peers. Through offering critical insights into the unique needs of international students dealing with the trauma of crises in their home countries, the study seeks to offer concrete insights about how to foster a humanistic, supportive environment that fosters their emotional well-being and academic success
STING: A Stealthy Backdoor Attack on GNN-Based Malicious Domain Detection via DNS Perturbations
No full textDetecting malicious Internet domains is essential for safeguarding against various online threats. The current approach to detecting malicious domains (MDD) employs a graph neural network (GNN) method, which uses DNS logs to construct heterogeneous graphs for determining the maliciousness of unknown domains. Despite its success, this method is vulnerable to data poisoning attacks where an adversary can manipulate specific graph nodes to implant a backdoor into the model during training. To showcase the vulnerability, we propose a stealthy trigger injection attack on node features and graph structure in MDD, dubbed (STING). The attacker carefully manipulates selected features and edges of its nodes in the graph to create backdoor trigger patterns designed to evade detection by the MDD system, without knowing the model or other parts of the graph. Results from experiments conducted on real-world GNN-based MDD approaches show that the proposed attack is highly effective, with a success rate of over 88% in launching backdoor attacks and only a slight decrease in the model’s accuracy on legitimate domains (not exceeding 4%). Furthermore, the attack bypasses established defenses such as graph purification, adversarial training, and outlier detection, making it a major threat to the security of MDD systems. This study serves as a warning and stresses the importance of continuous vigilance and proactive efforts by both researchers and security experts to secure GNN-based MDD systems and maintain their trustworthiness and stability
From Biophysics to Cellular Function: Neural TELCs-Membrane-Anions Capacitor Transmembrane Potential
No full textBased on the transmembrane-electrostatically localized protons/cations charges (TELCs) theory, neural transmembrane potential including both resting and action potential is now well elucidated as the voltage contributed by the TELCs-membrane-anions capacitor biophysics in a neuron. Accordingly, neural transmembrane potential has an inverse relationship with TELCs surface density, which may represent a substantial progress in bettering the fundamental understanding of neuroscience. In this article, I will present a review on the latest development of the TELCs neural transmembrane potential theory and address Silverstein’s interesting arguments regarding the TELCs model that may constitute a complementary development to both the Hodgkin-Huxley classic cable theory and the Goldman-Hodgkin-Katz equation. A series of predictions from the TELCs model regarding crucial ion channels have exactly been experimentally observed in many well-established electrophysiological phenomena including (but not limited to): 1) The tetrodotoxin (TTX) sensitivity shows the complete blockade of action potentials by TTX; 2) Genetic knockout or mutation of critical ion channels abolishes action potential spike; and 3) The precise clustering of ion channels at the axonal initial segment and nodes of Ranvier underlies the ability to fire action potential spikes and the saltatory conduction along a myelinated axon. This indicates that the TELCs model can be well predictive and provide new opportunities as a theoretical tool for further research to better understand neurosciences
Factorization for Jet Production in Heavy-Ion Collisions
No full textWe develop an Effective Field Theory approach for jet observables in heavy-ion collisions, where the jet is treated as an open quantum system interacting with a hot and dense QCD medium. Within this framework, we derive a novel factorization formula for inclusive jet production, expressed as a series expansion with an increasing number of radiating subjet functions that encode forward scattering with the Quark-Gluon Plasma, convolved with perturbative matching coefficients. This work provides a systematic framework for computing jet observables at higher order and understanding their non-perturbative aspects, paving the way for future applications in heavy-ion phenomenology
Dissociation of Subjective and Objective Measures of Trust in Vehicle Automation: A Driving Simulator Study
No full textTrust is a crucial factor that influences human-automation interaction in surface transportation. Previous research indicates that participants tend to display higher levels of subjective trust toward lower-level automated systems compared to high-level automated systems. However, administering subjective trust measures via questionnaires can interfere with primary task performance, limiting researchers\u27 ability to measure trust continuously in a real-world manner. In the current driving simulator study, 25 drivers using an advanced driving system (ADS) were randomly assigned to either an active (L2) or passive (L3) automated driving condition. Participants experienced eight near-miss driving scenarios with or without obstructions in a distributed driving simulator and rated their subjective trust before and after navigating the scenarios. Additionally, we coded hand positions from recorded video footage of the participants\u27 in-vehicle behavior. Hand placements were coded on a predefined five-point system near the time of the simulated connected vehicle technology\u27s collision alert. Results showed that drivers progressively lost trust in the automated system as they approached and passed the projected collision point in each scenario. Furthermore, drivers in the active condition displayed lower levels of trust than those in the passive condition. This finding contrasts with previous research suggesting that subjective trust ratings are comparable between Level 2 and Level 3 vehicle automation groups. These findings highlight a dissociation between subjective and behavioral measures of trust, suggesting that self-report methods may overlook important aspects of drivers\u27 trust that can be captured through behavioral measures
Food Insecurity and Academic Self-Efficacy: Military Students and Student Veterans
No full textFood insecurity (FI) affects households worldwide and is also seen on college and university campuses with adverse effects on student success. We examine FI and academic self-efficacy levels of military students and student veterans enrolled in higher education in the United States which has not been extensively studied regarding FI. Military students and student veterans may have access to G.I. Bill funds and may not be perceived as requiring additional assistance. The results reveal that military students and student veterans have some level of FI. The qualitative analysis yielded five themes that center on FI, economic vulnerability, inequities in the veteran community, reintegration challenges, and support system gaps