University of Illinois Urbana-Champaign
Illinois Digital Environment for Access to Learning and Scholarship RepositoryNot a member yet
123813 research outputs found
Sort by
Three essays on financial market networks
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Belinda Chen, accepted the attached license on 2025-04-07 at 10:53.The student, Belinda Chen, submitted this Dissertation for approval on 2025-04-07 at 10:57.This Dissertation was approved for publication on 2025-04-08 at 11:10.DSpace SAF Submission Ingestion Package generated from Vireo submission #21719 on 2025-10-19 at 19:52:47This dissertation explores the growing importance of network structures in financial markets by examining how firm-level interconnections shape risk transmission and aggregate outcomes. Through three essays, it highlights the role of production-based input-output networks in amplifying idiosyncratic shocks and driving both aggregate volatility and asset pricing dynamics. The first essay develops a dynamic model linking firm-level volatility spillovers to market-wide uncertainty, introducing novel network-based risk factors that are both predictive and priced. The second essay applies Graph Neural Networks to firm credit risk prediction, demonstrating how incorporating inter-firm network features enhances predictive power and interpretability. The final essay provides a theoretical foundation for how persistent, interconnected firm-level risks can generate macroeconomic tail events, even in the absence of large individual shocks. Together, these studies underscore the critical role of financial market networks in understanding modern economic and financial phenomena
Whataboutism with foreign affairs: How Chinese media leverage foreign news for distraction and legitimation
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Shuyuan Shen, accepted the attached license on 2025-04-17 at 23:18.The student, Shuyuan Shen, submitted this Dissertation for approval on 2025-04-17 at 23:35.This Dissertation was approved for publication on 2025-04-18 at 16:00.DSpace SAF Submission Ingestion Package generated from Vireo submission #21828 on 2025-10-19 at 19:53:20This dissertation examines the strategic use of foreign news as a propaganda tool in authoritarian regimes, focusing on China. While prior research emphasizes domestic narratives in authoritarian propaganda, this study highlights foreign news as a distinct form of propaganda for distraction and legitimation. Analyzing approximately two million social media posts from twelve major Chinese media outlets between 2011 and 2023, the dissertation first reveals that Chinese foreign news coverage focuses heavily on Western democracies, regional rivals, and the Middle East. Geographic and economic proximities increase a country’s media visibility, while cultural and political distances attract greater attention. The extensive coverage of foreign rivals rather than politically proximate countries as existing literature suggests indicates potential strategic intent. The dissertation then demonstrates that negative foreign news is strategically employed during economic downturns to distract public focus from domestic concerns. Statistical analyses link rising inflation and public attention to inflation and unemployment to increased negative coverage of foreign economic and political affairs. A pre-registered survey experiment confirms that negative foreign news effectively shifts attention away from domestic issues, outperforming positive domestic propaganda as a distraction instrument. Moreover, negative foreign news reinforces authoritarian legitimacy by exploiting national identity, anti-foreign predispositions, and cognitive negativity biases. Experimental and observational evidence confirms that such coverage undermines perceptions of foreign democracies while bolstering domestic regime support. Finally, the dissertation extends the study of foreign-focused information to foreign political entertainment, revealing that its influence on public perceptions is modest and inconsistent with a second survey experiment in China. By integrating insights from communication studies, political psychology, and authoritarian politics, this research expands the understanding of propaganda strategies and the political influence of foreign-focused information in non-democratic contexts
Improving neuron-level interpretability with white-box language models
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Hao Bai, accepted the attached license on 2025-04-24 at 10:25.The student, Hao Bai, submitted this Thesis for approval on 2025-04-24 at 10:31.This Thesis was approved for publication on 2025-04-24 at 17:11.DSpace SAF Submission Ingestion Package generated from Vireo submission #21944 on 2025-10-19 at 19:53:52Neurons in auto-regressive language models like GPT-2 can be interpreted by analyzing their activation patterns. Recent studies have shown that techniques such as dictionary learning, a form of post-hoc sparse coding, enhance this neuron-level interpretability. In our research, we are driven by the goal to fundamentally improve neural network interpretability by embedding sparse coding directly within the model architecture, rather than applying it as an afterthought. In our study, we introduce a white-box transformer-like architecture named Coding RAte TransformEr (crate), explicitly engineered to capture sparse, lowdimensional structures within data distributions. Our comprehensive experiments showcase significant improvements (up to 103% relative improvement) in neuron-level interpretability across a variety of evaluation metrics. Detailed investigations confirm that this enhanced interpretability is steady across different layers irrespective of the model size, underlining crate’s robust performance in enhancing neural network interpretability. Further analysis shows that crate’s increased interpretability comes from its enhanced ability to consistently and distinctively activate on relevant tokens. These findings point towards a promising direction for creating white-box foundation models that excel in neuron-level interpretation
In the name of democracy: Composition, variation, and measurement of conceptions of democracy
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Seongjoon Ahn, accepted the attached license on 2025-04-24 at 12:58.The student, Seongjoon Ahn, submitted this Dissertation for approval on 2025-04-24 at 12:58.This Dissertation was approved for publication on 2025-04-24 at 16:58.DSpace SAF Submission Ingestion Package generated from Vireo submission #21954 on 2025-10-19 at 19:54:00In what ways do conceptions of democracy differ among Americans? Even though much of the empirical literature in political science seems to assume a singular conception of democracy, ordinary Americans differ in how they conceive of what democracy is. Both the participants of the January 6th Attack on the Capitol and the general public point to “saving democracy” as a driving motivation, but they exhibit drastically different behaviors. In this dissertation, I conduct a comprehensive investigation of Americans’ conceptions of democracy. I first develop a theoretical framework to examine how individuals conceptualize democracy and how these conceptions relate to political behavior. I then test and apply this framework using two empirical approaches. In Chapter 4, I employ mixed-methods analysis to explore variation in democratic conceptions among a small group of study participants. In Chapter 5, I use latent class analysis on an original, state-level representative survey of the American public (N = 25,902). In Chapter 6, I examine the factors driving these conceptual differences, identifying the demographic and attitudinal traits associated with particular understandings of democracy. In Chapter 7, I test the relationship between individuals’ conceptions of democracy and various political attitudes and find that, even after controlling for major demographic variables, variations in conceptions of democracy are strongly correlated with variations in political attitudes. The results show that 1) five different conceptions of democracy compete within the American public, 2) the division cuts across traditional sociopolitical cleavages, 3) certain conceptions deviate from the more conventionally accepted understandings of democracy, and 4) the variations in conceptions of democracy are highly correlated with variations in political attitudes. These findings suggest that democracy is understood as a multidimensional concept and that people hold distinct and sometimes contradicting composite conceptions of democracy. Furthermore, they imply that conflicting conceptions of democracy may help explain the conflicting attitudes of small-d democrats in the United States
Application of spectral techniques and machine learning for fertility, mortality, sex, and structural evaluation of eggs
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, MD Wadud Ahmed, accepted the attached license on 2025-04-29 at 11:40.The student, MD Wadud Ahmed, submitted this Dissertation for approval on 2025-04-29 at 11:52.This Dissertation was approved for publication on 2025-05-01 at 18:56.DSpace SAF Submission Ingestion Package generated from Vireo submission #22078 on 2025-10-19 at 19:54:36Eggs are considered one of the best dietary protein sources and have an important role in human growth and development. Fast and accurate quality assessment and grading of eggs is crucial to meet the growing demand for high-quality eggs driven by rapid population growth, economic sustainability, and increasing animal welfare concerns. Evaluation of some key parameters, such as fertility, mortality, sex, shell thickness, shell strength, and yolk ratio (i.e., yolk to egg mass ratio), is essential for optimizing hatchery productivity, ensuring egg quality, and maintaining ethical standards in the egg industry. Conventional methods for evaluating these parameters, such as candling, manual measurement, and mechanical testing, are inherently labor-intensive, time-consuming, and destructive, highlighting the need for non-destructive techniques in the egg industry. Spectral techniques such as near-infrared (NIR) spectroscopy and hyperspectral imaging (HSI) offer rapid, non-invasive, and accurate alternatives to conventional techniques, enabling the simultaneous assessment of multiple parameters without compromising the integrity of eggs. Spectral techniques collect spectra at numerous narrow and contiguous wavelengths. Consequently, chemometrics and machine learning (ML) techniques facilitate the analysis of these spectra by identifying complex patterns and relationships in the data, enabling accurate prediction, classification, and evaluation of egg parameters. The objective of this research was to apply spectral techniques (NIR spectroscopy and HSI) and ML for fertility, mortality, sex, shell thickness, shell strength, and yolk ratio analysis of eggs. This study can enhance automation, improve quality control, promote animal welfare, and optimize resource management for the sustainable development of the table and hatching egg industry. The poultry industry heavily relies on accurate detection of egg fertility to optimize hatchery operations. In this study, HSI spectra was used to develop pre-incubation chicken egg fertility classification models using XGBoost, CatBoost, RF, and SVM. Using full wavelengths, the CatBoost model with synthetic data showed the best classification performance, attaining 95.1% accuracy in independent validation. The CatBoost models with fewer important features showed good prediction performance, making them computationally efficient, robust, and interpretable. The Shapley additive explanation (SHAP) explainable AI technique was used to interpret the robust CatBoost model, revealing that wavelength regions associated with yolk color, pre-incubation cellular activities related to embryonic development, changes in hydration levels, and variations in protein and lipid contents between fertile and infertile eggs are crucial for pre-incubation chicken egg fertility classification. Non-destructive pre-incubation and early incubation chick embryo mortality prediction is crucial for optimizing hatchery operations and improving poultry production. This study explores potential of visible-near infrared (Vis-NIR) HSI combined with ML models to classify embryo mortality during pre-incubation and at 4 days of incubation periods. Partial least squares discriminant analysis (PLS-DA), RF, and CatBoost calibration models were developed, and performance of the calibration models was evaluated by independent validation and test sets. At full wavelength (501-921 nm), the PLS-DA model demonstrated the best performance for chick embryo mortality classification, achieving an accuracy of 91.3% in calibration, 88% in validation, and 86.7% in the test set for pre-incubation, while for day 4 incubation, it attained 97.3% accuracy in calibration, 96% in validation, and 97.3% in the test set, highlighting its robustness across different data sets. PLS-DA models with a reduced set of important spectral features demonstrated strong predictive performance, offering computational efficiency, robustness, and enhanced interpretability. SHAP model explantion revealed that wavelengths related to blood formation, embryo hydration status, and metabolic variations between live and dead embryos are critical for chick embryo mortality classification during early incubation. Non-destructive sex determination in eggs can enhance animal welfare, improve economic efficiency, reduce environmental impact, and foster technological innovation in sustainable hatchery operations. This study developed different classification models for pre-incubation sex prediction in chicken egg using HSI and ML. Partial least squares discriminant analysis (PLS-DA), XGBoost, RF, and CatBoost classification models were constructed using full-wavelength (452-899 nm) spectra and evaluated them through external validation. Multiple spectral pre-processing methods, feature selection approaches, and hyperparameter tuning were assessed to optimize and develop robust prediction models. Full featured CatBoost model exhibited superior performance among models tested, with 88.6% calibration accuracy and 75.5% validation accuracy. With 35 important features based optimized CatBoost model showed promising results with an accuracy of 82.9% and 75.5% in calibration and independent validation data. The eggshell protects internal contents, supplies calcium for embryos, and aids embryonic respiration. This study assessed NIR spectroscopy, ML, and explainable AI for real-time eggshell thickness prediction by developing partial least squares regression (PLSR), RF, K-nearest neighbors (KNN), and support vector regression (SVR) models using full wavelengths (1300-2525 nm) and selected important wavelengths. The PLSR model demonstrated superior and stable predictive performance, achieving a coefficient of determination (R2p) of 0.867, and root mean square error of prediction (RMSEP) of 0.015 mm. A new PLSR model using only five important variables showed promising results with an R2p of 0.910 and RMSEP of 0.012 mm. The SHAP explanation of the final PLSR model revealed wavelengths related to protein, moisture, and lipids are crucial for NIR spectroscopic prediction of eggshell thickness. Eggshell strength is crucial for ensuring high-quality eggs, reducing breakage during handling, and meeting consumer expectations for freshness and integrity. This study evaluated potential of NIR spectroscopy (1300-2525 nm), ML, and SHAP explainable AI for rapid, non-destructive method for detecting eggshell strength. Principal component analysis (PCA) and PLS-DA effectively classified eggs based on a threshold shell strength of 30 N. Regression models, including PLSR, RF, Light GBM, and KNN were used to predict eggshell strength and evaluated by independent validation test. Using only 14 selected important variables, the RF model achieved a very good prediction performance with coefficient of determination of prediction (R2p) of 0.83, minimum RMSEP of 1.49 N, and RPD of 2.44. The SHAP explanation of the RF model revealed wavelengths related to moisture (of shell and albumen) were crucial for NIR spectroscopic prediction of eggshell strength. In the final part, the potential of HSI combined with ML and explainable AI for yolk ratio prediction was investigated. Multiple full spectral (374-1015 nm) regression models were developed using the calibration set, validated with an independent dataset, and further tested using another independent dataset, ensuring robustness and generalizability across varying samples. The PLSR model showed superior predictive performance (R2: calibration=0.79, validation=0.73, test=0.68), with similarly robust results obtained using only 20 important wavelengths. SHAP explanation of the final PLSR model revealed wavelength region associated with water had the most influence on non-destructive yolk ratio prediction by HSI
How do firms recognize and present their human capital?: investigation of HR executive appointment and human capital disclosures
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Suyeon Kang, accepted the attached license on 2025-04-30 at 13:06.The student, Suyeon Kang, submitted this Dissertation for approval on 2025-04-30 at 13:09.This Dissertation was approved for publication on 2025-05-01 at 14:56.DSpace SAF Submission Ingestion Package generated from Vireo submission #22117 on 2025-10-19 at 19:54:49This dissertation examines how U.S. corporations recognize and present human capital through three complementary studies. The first chapter investigates the rise of Human Resources (HR) executives to top management teams, revealing that, contrary to strategic human resource rhetoric, this elevation is primarily driven by administrative needs—such as workforce downsizing and union density—rather than a recognition of HR's strategic value. The second chapter analyzes how different types of institutional investors influence human capital (HC) disclosure compliance following the SEC’s 2020 HC disclosure mandate. The results demonstrate that long-term dedicated investors positively affect disclosure comprehensiveness, while short-term-oriented investors exert a negative influence, with quasi-indexers shifting from indifference to positive influence after the regulatory change. The third chapter explores mimetic isomorphism in human capital disclosures using natural language processing and network analysis. It demonstrates that isomorphism develops in a non-linear manner, resulting in globally dispersed but locally concentrated network formations. Network core analysis reveals that smaller, innovation-intensive firms establish influential disclosure templates under regulatory uncertainty. Together, these studies demonstrate how competing organizational priorities shape corporate human capital practices in response to external and stakeholder pressures. The findings provide a comprehensive understanding of how firms recognize, manage, and present human capital within evolving institutional and regulatory environments
Integrated anaerobic digestion and electrodialysis process model for enhanced volatile fatty acids recovery from thin stillage
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Junhyung Park, accepted the attached license on 2025-05-03 at 02:51.The student, Junhyung Park, submitted this Thesis for approval on 2025-05-03 at 03:30.This Thesis was approved for publication on 2025-05-09 at 11:29.DSpace SAF Submission Ingestion Package generated from Vireo submission #22186 on 2025-10-19 at 19:55:42Anaerobic digestion (AD) is widely applied for renewable energy recovery via biogas production. However, due to the abundant availability and low price of natural gas, biogas as a renewable energy source faces economic limitations. To enhance the sustainability and economic viability of AD, recent efforts have focused on uncoupling methanogenesis to produce valuable carboxylates, especially volatile fatty acids (VFAs). Efficient separation and recovery of these VFAs represent significant technological and economic challenges. This study focuses on modeling and optimizing integrated AD and electrodialysis (ED) processes for enhanced VFA production and separation. The Anaerobic Digestion Model No. 1 (ADM1) was modified within the QSDsan platform to accurately represent fermentation pathways, including lactate and ethanol formation. Experimental data obtained from continuous operation of hybrid bioreactors inoculated with cow manure, using glucose and thin stillage as substrates, were used to calibrate the model. Simulation results demonstrated the critical influence of operational pH on the metabolic pathways, promoting either biogas or VFA production. An ED process model developed in BioSTEAM allowed for the evaluation and optimization of key operational parameters such as membrane area, current density, and hydraulic retention time under steady-state and continuous conditions. Techno-economic analysis (TEA) revealed membrane area and current density as critical cost determinants, with optimized conditions significantly reducing energy consumption and operating costs. Overall, this integrated modeling approach provides valuable insights into the scalability and economic feasibility of sustainable carboxylate recovery processes, offering a competitive alternative to traditional separation methods such as multi-effect evaporation
Evaluation of the integration of emerging sanitation technologies in centralized wastewater treatment systems
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Nehal Jain, accepted the attached license on 2025-05-06 at 21:04.The student, Nehal Jain, submitted this Thesis for approval on 2025-05-06 at 21:09.This Thesis was approved for publication on 2025-05-08 at 15:56.DSpace SAF Submission Ingestion Package generated from Vireo submission #22241 on 2025-10-19 at 19:55:48Centralized wastewater infrastructure in the U.S. faces critical challenges - aging systems, chronic underinvestment, operational overloading, and climate-induced vulnerabilities - that impair treatment efficacy and lead to regulatory non-compliance and environmental externalities. Integration of emerging non-sewered sanitation (NSS) technologies with traditional treatment facilities offer systemic improvements, enhance resilience and sustainability of wastewater management. Source separation of urine is one such NSS technology that reduces nutrient loads to centralized water resource recovery facilities (WRRFs), directly influencing the operational costs and energy demands. It is therefore essential to benchmark integrated urine diversion (UD) systems against conventional wastewater treatment processes. The primary aim of this study is to investigate how upstream urine diversion influences the key operational and environmental metrics – namely, costs and greenhouse gas (GHG) emissions – at centralized WRRFs. The modeled WRRF is a representative configuration targeting biological nutrient removal (BNR). Modified versions of Activated Sludge Model No. 2D and Anaerobic Digestion Model No. 1 (mASM2d and ADM1p, respectively), which include phosphorus (P) transformations, are used to model the biological processes. Plant-wide P dynamics were modeled using interfaces that transition between the two process models. QSDsan, an open-source, community-led, Python-based platform tailored to the quantitative sustainable design (QSD) of sanitation and resource recovery systems is used to model the WRRF configuration coupled with UD. Monte Carlo simulations were performed to evaluate two scenarios - conventional (without UD) and with full UD - under uncertainty. Results show that integrating UD leads to significant reductions in operational costs (~27%) and GHG emissions (~38%) at BNR facilities. The key drivers of operational costs and GHG emissions were identified by obtaining the Spearman’s rank correlation coefficients of these metrics with the continuously varying contextual parameters. Result of the sensitivity analysis revealed that operational costs were primarily influenced by the unit cost of electricity, which also remained the most significant driver of cost reductions following urine diversion. In parallel, the emission factor (EF) for N2O during secondary treatment was the key determinant of GHG emissions and continued to dominate the sensitivity of emission reductions post-diversion
Comparison of grassland bird stopover and breeding habitat use in Illinois
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Nicole Suckow, accepted the attached license on 2025-05-07 at 12:33.The student, Nicole Suckow, submitted this Thesis for approval on 2025-05-07 at 12:43.This Thesis was approved for publication on 2025-05-08 at 08:48.DSpace SAF Submission Ingestion Package generated from Vireo submission #22251 on 2025-10-19 at 19:55:49Grassland birds are declining at an alarming rate, and there has been little research on the stopover behavior and habitat needs of migratory grassland species. Considering heightened mortality risk during migration, uncovering what attributes comprise good stopover habitat is necessary to best conserve these species. My thesis research is focused on the habitat use of grassland birds listed as Species of Greatest Conservation Need (SGCN) in Illinois, especially during spring and fall migration. To find which grassland characteristics were most important to each species, I surveyed grassland birds from late-March/early-April through October in 2022 and 2023. During this period, I also collected data on vegetation and management activities. Logistic regression models were used to analyze and compare the impact of vegetation composition, type, and density, as well as site size, management activities, and surrounding landscape on the likelihood of each species’ presence. The habitat variables that best explained grassland bird detections at any given site were the surrounding landscape and vegetation composition. However, the variable(s) influencing the presence of grassland birds differed between species and season. Conservation efforts could target individual species by focusing on localized management to attain preferred vegetation structure, but the biggest impact for grassland SGCN as a whole would be to prioritize creating landscapes that contain more grasslands
A computer vision-based dimension measurement method for visual inspection system in smart manufacturing
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Shitao Liu, accepted the attached license on 2025-05-09 at 16:53.The student, Shitao Liu, submitted this Thesis for approval on 2025-05-09 at 16:56.This Thesis was approved for publication on 2025-05-09 at 17:02.DSpace SAF Submission Ingestion Package generated from Vireo submission #22293 on 2025-10-19 at 19:55:54Accurate dimension inspection is crucial in manufacturing. The proper functionality of the manufactured products is premised on the accurate dimension recognization, localization, and measurement. Traditionally, this task can be conducted by a trained technician. However, as the global electronics market grows and the development of semi-conductor industry progresses, the manufactured products become smaller in size but larger in quantity, which poses a significant challenge to the manufacturers. Thus, developing and deploying an automated inspection system with high accuracy and fidelity is needed to ensure the smooth operation of these manufacturers. In this thesis, I develop a dimension inspection system based on applying computer vision techniques to analyzing the photos of manufactured parts to solve this question. With CAD drawings and sample part images provided by Foxconn Interconnect Technology (FIT), I design a workflow to facilitate the inspection process, from critical feature recognition to data reporting. The backbone of this inspection system is based on image-level signal filtering and geometry detection to ensure the flexibility and generalizability across different CAD designs. The performance of this inspection system was tested on a large dataset of six different dimensions from the USB-C connector FIT manufactures, as well as our proprietary 3D printing dataset, and the result shows the industrial application potential of this system