University of Illinois Urbana-Champaign
Illinois Digital Environment for Access to Learning and Scholarship RepositoryNot a member yet
123813 research outputs found
Sort by
A GUIDE TO ROBERTO PIANA’S 25 PRELUDI PITTORICI
No full textThis essay explores the life and musical contributions of Italian composer and pianist Roberto Piana, with a particular focus on his cycle 25 Preludi Pittorici. Focusing on twenty-five paintings from the eighteenth to twentieth centuries, Piana creates a musical gallery that transforms
art into sound.
The study first contextualizes Piana’s career and artistic influences. An analysis then examines the correlation between selected paintings and their corresponding preludes, highlighting how musical gestures reflect visual characteristics and narrative atmosphere. Attention is also given to the symbolic role of the sea, a recurring element tied to the composer’s Sardinian identity, and his
use of symbolic structures such as the golden ratio.
Through this synthesis of biography, analysis, the essay situates Piana’s Preludi Pittorici within the broader tradition of music inspired by the visual arts, while underscoring its unique contemporary voice
Enhancing the Transition into Adulthood for Foster Youth: Analysis of Illinois House Bill 1293
No full textThis paper explores the transition of foster youth into adulthood, focusing specifically on the challenges during the “aging out” process. Illinois House Bill 1293 extends foster care services until age 23, aiming to improve outcomes. However, the policy lacks clarity on self-sufficiency criteria, hindering effective implementation. Recommendations include comprehensively defining self-sufficiency and empowering youth to advocate for extended support– if desired or necessary. Tailored service planning is essential to address individual needs and prevent adverse outcomes. While Illinois House Bill 1293 is a step forward, clarity, accountability, and targeted support are crucial for successful transitions into adulthood for foster youth
Characterization and control of deployable origami structure towards a sustainable built environment
No full textSubmission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2027-05-01The student, Angshuman Baruah, accepted the attached license on 2025-04-23 at 14:14.The student, Angshuman Baruah, submitted this Dissertation for approval on 2025-04-23 at 14:25.This Dissertation was approved for publication on 2025-04-25 at 16:39.DSpace SAF Submission Ingestion Package generated from Vireo submission #21922 on 2025-10-19 at 19:16:01The thesis investigates the design, characterization, and control of deployable origami structures as innovative and sustainable solutions for infrastructure. Employing biomimetics to replicate the defensive conglobation behavior of pill bugs and leveraging principles of origami mechanics, a novel modular structure called the Origami Pill Bug is developed. The Origami Pill Bug features a plate-based deployable system capable of transitioning between flat and rolled configurations, offering promising applications in emergency shelters and adaptive civil engineering structures. A hybrid approach combining computational analysis and experimental studies is employed to investigate the structural behavior of the Origami Pill Bug. A bar-and-hinge approximation combined with dynamic relaxation is developed to accurately model the Origami Pill Bug’s nonlinear geometric transformations. This modeling approach facilitates the form-finding of deployment shapes, which are subsequently used to generate finite element models for modal analyses. Multiple prototypes are developed and refined, culminating in the construction of the final meter-scale prototype. This meter-scale prototype is then experimentally tested to validate computational predictions of natural frequency variations during deployment. The comparison confirms the robustness of the proposed hybrid modeling approach. Moving towards structural health monitoring, a multi-objective optimization framework is employed for optimal sensor placement. Experimental investigations determine efficient actuation rates for deployment, achieving a balance between operational speed and structural integrity. This research further advances damage detection capabilities through supervised machine learning algorithms, successfully classifying multiple damage scenarios using strain profile data. The environmental impact of the Origami Pill Bug is evaluated through a comparative life cycle assessment against traditional emergency shelter structures. The results highlight the Origami Pill Bug’s potential as a sustainable alternative to traditional shelters, emphasizing its adaptability and reduced ecological footprint. This research addresses critical challenges in scalability, dynamic performance, and environmental impact assessment of deployable origami structures, contributing novel insights to the field. The findings have significant implications for disaster response, modular construction, and the design of resilient infrastructure, paving the way for more adaptable and sustainable built environments
High-speed signal integrity analysis and channel modeling using neural networks
No full textSubmission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2027-05-01The student, Juhitha Konduru, accepted the attached license on 2025-05-01 at 05:45.The student, Juhitha Konduru, submitted this Dissertation for approval on 2025-05-01 at 06:03.This Dissertation was approved for publication on 2025-05-02 at 12:10.DSpace SAF Submission Ingestion Package generated from Vireo submission #22143 on 2025-10-19 at 19:16:56The analysis of high-speed networks is often carried out using transistor-level simulation tools which have large computational time. This leads to a limitation in terms of the amount of time spent generating an optimal design and accurately analyzing the system. Therefore, there is a need for fast and accurate modeling of packages and boards, which is the key for developing high performance devices. With increasing complexity, thermal effects significantly impact the systems performance as well. Hence, the fast model should be able to perform electro-thermal co-simulations as well. By integrating thermal analysis with electrical simulations, we can optimize designs for efficiency without overheating issues. This thesis discusses a machine learning based approach using neural networks to generate a fast model, eliminating the need to run long simulations using EM solvers often. This helps in creating the optimal design faster without going through many iterations. An ML-based fast-learned model is obtained for a differential PTH. An effective way to generate datasets for training the ML model is discussed. The generated ML model shows a 200X improvement over HFSS while simulating a single design using the Inference model of the neural network. This thesis also discusses a method using machine learning to perform electro-thermal simulations. The proposed method shows a 220X speedup when compared to the two-way coupling process for electro-thermal simulations
Development of rotary reactor for evaluation of feedstock mixing effect on hydrothermal liquefaction
No full textSubmission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2027-05-01The student, Koji Okutomi, accepted the attached license on 2025-05-02 at 11:49.The student, Koji Okutomi, submitted this Thesis for approval on 2025-05-02 at 11:51.This Thesis was approved for publication on 2025-05-09 at 13:42.DSpace SAF Submission Ingestion Package generated from Vireo submission #22178 on 2025-10-19 at 19:17:00Hydrothermal liquefaction (HTL) uses high temperature and pressure to convert biomass into biocrude and other byproducts. HTL mimics similar conditions that occur below ground on Earth which makes crude oil from biomass over millions of years. With HTL processes one can make biocrude oil in minutes, which explains its growing area of interest today. This is a thermochemical process in a family of other hydrothermal processes including hydrothermal gasification (HTG) and hydrothermal carbonization (HTC), Elhassan et al. (2023). HTC typically occurs at lower temperatures of 180 – 300 ᵒC and lower pressures of 2 – 6 MPa. The main product of hydrothermal carbonization (HTC) is biochar, and it is a similar process to how coal naturally forms. On the other end of the spectrum, HTG occurs at higher temperatures of above 374 ᵒC and a pressure of 22 MPa. HTL falls in between at approximate temperatures of 250 – 374 ᵒC and pressures between 4 – 22 MPa. HTL uses water as a solvent and reactant in the process. Current lab-scale tubular HTL reactors are widely used for optimizing HTL conditions. It is advantageous over the larger batch reactors because it is easy to operate and quick data generation. However, tubular HTL reactors typically operate in a stationary position without mixing of the feedstock. Mixing in a hydrothermal liquefaction reactor could be influential to the HTL efficiency and quality of the product. By using rotation, the heat and mass transfer in the HTL reactor can be improved thus enhancing the HTL conversion efficiency and product quality. In this project, a rotational mechanism was designed, fabricated, and integrated into a heating furnace to establish a rotary HTL reactor system. The system is a lab scale system with approximate dimensions of 1.02 x 0.53 x 0.38 m. This new HTL reactor enhances greatly HTL research by allowing multiple reactions to be used simultaneously under different mixing conditions. Initial experiments were conducted to validate the system as well as test the mixing effects
Hierarchical modeling of systemic risk via mean field games
No full textSubmission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2027-05-01The student, Prathmesh Rathod, accepted the attached license on 2025-05-05 at 16:08.The student, Prathmesh Rathod, submitted this Thesis for approval on 2025-05-05 at 16:09.This Thesis was approved for publication on 2025-05-08 at 17:41.DSpace SAF Submission Ingestion Package generated from Vireo submission #22218 on 2025-10-19 at 19:17:03In the aftermath of the 2008 global financial crisis, the question of how to manage and mitigate the systemic risk in the banking sector has become a cornerstone of financial regulation objectives and policy. Motivated by this observation, this thesis aims to model the mitigation of cascading systemic risk at the local bank level by introducing the policies of the central banks and International Monetary Fund (IMF) into the modeling. Specifically, we introduce a mathematical model in which local banks in K many countries, their central banks, the IMF is modeled in a game theoretical setup. In order to model the game problem among large number of local banks in each country, we use the mean field game (MFG) methodology. We accomplish this by extending the model of Carmona, Fouque, and Sun [1] to the case where we can accommodate multiple populations that represent different countries. Therefore, we first give the mathematical model of the local banks, define the multi-population MFG Nash equilibrium for them, and present the theoretical characterization results by using Pontryagin maximum principle given the policies of the central bank and the IMF. Later, we introduce the mathematical model of the central banks and we present the Nash equilibrium definition between the central banks and local banks. We conclude by introducing the characterization result for the Nash equilibrium between central banks and local banks by using forward backward stochastic differential equations. Finally, we introduce the model of IMF and define the Stackelberg equilibrium in the whole system where IMF sets country specific policies and optimize these policies by taking into account the Nash equilibrium response of the central and local banks in each country
Harnessing AI/ML for advancing synthetic biology
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Tianhao Yu, accepted the attached license on 2025-04-29 at 13:44.The student, Tianhao Yu, submitted this Dissertation for approval on 2025-04-29 at 13:51.This Dissertation was approved for publication on 2025-04-30 at 09:05.DSpace SAF Submission Ingestion Package generated from Vireo submission #21323 on 2025-10-19 at 19:52:07The intersection of synthetic biology and artificial intelligence has created unique opportunities for understanding and engineering biological systems. This dissertation presents a suite of machine learning (ML) approaches tailored for protein and small molecule discovery, with an emphasis on integrating biological data and experimental validation. Spanning enzyme annotation, protein engineering, antimicrobial compound discovery, and multimodal representation learning, the research showcases how modern ML/AI methods can overcome key bottlenecks in synthetic biology and accelerate the design–build–test–learn cycle. Chapter 1 introduces the foundational concepts of synthetic biology, protein engineering, and ML. It highlights the recent proliferation of protein language models and their ability to extract biologically meaningful patterns from sequence data. The chapter also outlines the four major research threads explored in the thesis: enzyme function prediction using contrastive learning, ML-guided protein engineering, large language model-driven antimicrobial molecule discovery, and multimodal representation learning for enzyme annotation. Chapter 2 describes the development of CLEAN (Contrastive Learning-enabled Enzyme Annotation), a deep learning framework for annotating enzyme functions based solely on amino acid sequences. CLEAN utilizes contrastive learning to create EC number-aware protein representations by training on functionally similar and dissimilar enzyme pairs. The model significantly outperforms traditional similarity-based annotation tools and other ML-based tools on benchmark datasets, especially in retrieving understudied or misannotated enzymes. CLEAN also demonstrated practical utility through the experimental validation of novel halogenase candidates, uncovering a promiscuous enzyme with three catalytic activities. Chapter 3 presents a generalized ML framework for protein engineering, integrating zero-shot language model predictions with supervised learning in a closed-loop optimization pipeline. The chapter also introduces ECNet, an evolutionary context-integrated neural network trained to predict variant fitness using homologous sequence data. Further, this chapter describes a fully autonomous protein engineering platform combining ML as decision maker and robotics as experimentalists. This system was used to engineer two industrially relevant enzymes: AtHMT, to enhance ethyltransferase activity for S-adenosylmethionine analog synthesis, and YmPhytase, to broaden pH tolerance for animal feed applications. Both enzymes underwent 4 rounds of prediction, testing, and re-training, exemplifying a robust ML-guided directed evolution loop. The chapter also introduces a GPT-based user interface for protein variant design, lowering the barrier for non-expert users. Chapter 4 shifts focus to small molecules, detailing an LLM-driven framework for predicting antimicrobial activity of compounds. Using a SMILES-based transformer model pretrained on PubChem, the study built a regression pipeline to predict the minimal inhibitory concentration (MIC) of molecules against 12 Gram-negative bacterial species. Screening compound libraries like ZINC, the pipeline led to the identification of Diamiquincin (DAQ), a novel compound with potent activity against Acinetobacter baumannii (MIC = 2 µg/mL). This chapter illustrates the effectiveness of predictive LLMs in streamlining antibiotic discovery. Chapter 5 explores multimodal contrastive learning to improve enzyme function annotation beyond primary sequence data. Recognizing that biological function can be augmented with multiple modality (e.g., literature annotations), the chapter develops a dual-encoder architecture inspired by CLIP. Protein sequences encoded by ESM-2 and textual descriptions encoded by BioGPT are aligned in a shared latent space using InfoNCE-style loss functions. The model demonstrated superior performance over unimodal baselines in EC number prediction on the challenging independent dataset. Additionally, the "sequence-first" loss configuration, balancing between alignment and model simplicity, proved more effective than a fully mixed loss. These findings highlight the power of weak supervision from unstructured text to enhance protein representations, especially in applications where experimental labels are limited. Across all chapters, the thesis emphasizes co-design between computational models and experimental workflows. Each ML method is tightly coupled with wet-lab validation, establishing trust in predictions and enabling iterative improvement. Thematically, the work advances three core capabilities in synthetic biology: (1) understanding protein function at scale, (2) engineering protein function efficiently, and (3) discovering therapeutic compounds in a data-driven manner. Moreover, the models developed in this dissertation are built for generalizability and extensibility, setting the stage for future integration into broader synthetic biology toolkits. In summary, this dissertation demonstrates the potential of AI and ML to accelerate discovery and design in synthetic biology. By combining contrastive learning, large language models, and evolutionary insights, the research addresses long-standing challenges in enzyme annotation, protein engineering, and antibiotic discovery. The approaches outlined here exemplify a new generation of data-driven biology, where machine learning not only interprets biological data but also shapes future experimentation
Investigation of electrically small antennas designed for directive radiation from transient signaling
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, David Mitchell, accepted the attached license on 2025-02-28 at 11:09.The student, David Mitchell, submitted this Dissertation for approval on 2025-02-28 at 11:17.This Dissertation was approved for publication on 2025-03-04 at 15:44.DSpace SAF Submission Ingestion Package generated from Vireo submission #21653 on 2025-10-19 at 19:52:23Electrically small antennas are a useful subset of antennas because of their small physical size; however, they are inherently limited in their performance because of the relation between electrical size and bandwidth. Many authors have recently sought to improve the performance of electrically small antennas by introducing nonlinear or time-varying components, using transient signaling, or increasing the directivity of the antenna. This work explores an antenna system that leverages the inherent narrowband nature of an electrically small antenna to achieve directive radiation from the transient ringing of the antenna system. This novel antenna is first explored through simulation. Next, a circuit model for predicting the operating frequency of this antenna is developed. Several versions of this antenna are then manufactured and characterized. Finally, a method for characterizing the transient behavior of an antenna is described and used to demonstrate the directive transient radiation of this antenna. This dissertation demonstrates the design of an antenna shown to have directive radiation with both continuous and transient excitations and methods for characterizing that radiation
Explainable artificial intelligence and deep reconstruction of hyperspectral images for advancing sweetpotato quality evaluation
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Md Toukir Ahmed, accepted the attached license on 2025-04-02 at 17:36.The student, Md Toukir Ahmed, submitted this Dissertation for approval on 2025-04-02 at 17:36.This Dissertation was approved for publication on 2025-04-07 at 10:25.DSpace SAF Submission Ingestion Package generated from Vireo submission #21710 on 2025-10-19 at 19:52:46Sweetpotato (Ipomoea batatas L.) is valued for its rich nutritional content, economic benefits, and versatility in both food and industrial applications, making its quality assessment essential. However, traditional methods for evaluating sweetpotato quality are often time-consuming and destructive. Hyperspectral imaging (HSI) has emerged as a powerful tool to assess internal composition and texture by capturing detailed spatial and spectral data. However, HSI generates high-dimensional data, which presents substantial computational challenges. To address these challenges, chemometrics is employed for effective data reduction and interpretation. In this context, explainable artificial intelligence (XAI) has become increasingly important in making predictive models more transparent and interpretable, thereby enhancing the reliability and adoption of artificial intelligence (AI)-driven quality assessments. Concurrently, advances in deep learning have opened new avenues for reconstructing hyperspectral data from standard RGB images, offering a more accessible and cost-effective alternative to traditional HSI systems. This approach, though underexplored in agricultural applications, holds significant potential for improving the practicality and efficiency of sweetpotato quality evaluation, making it a promising area for further research and development. This research aims to utilize the implementation of XAI and HSI reconstruction techniques to enhance the accuracy, transparency, and accessibility of sweetpotato quality assessment, ultimately contributing to the optimization of agricultural practices and industrial applications. In the first part of the study, XAI was integrated with hyperspectral imaging to enhance the assessment of three important quality attributes in sweetpotatoes, i.e., dry matter content (DMC), soluble solid content (SSC), and firmness. Sweetpotato samples of three different varieties, including “Bayou Belle”, “Murasaki”, and “Orleans”, were imaged using a portable visible near-infrared hyperspectral imaging (VNIR-HSI) camera, with a 400-1000 nm spectral range. The extracted spectral data were used to select key wavelengths, develop partial least squares regression (PLSR) models, and utilize shapley additive explanations (SHAP) values to ascertain model effectiveness and interpretability. The regression models (dry matter: R2p = 0.92, RMSEP = 1.50% and RPD = 5.58; soluble solid content: R2p = 0.66, RMSEP = 0.85obrix, and RPD =1.72; firmness: R2p = 0.85; RMSEP = 1.66N and RPD = 2.63) developed with key wavelengths were used to generate prediction maps to visualize the spatial distribution of response attributes, facilitating an improved evaluation of sweetpotato quality. Multivariate modelling techniques such as PLSR are commonly used for their simplicity, speed, and performance in industrial spectroscopic applications. While these models handle mild nonlinearities well, they often struggle with extrapolation. Therefore, the second part of the study aimed to utilize HSI and convolutional neural networks (CNN)-based regression to predict the firmness of various sweetpotato varieties by extracting spectral data from images captured with a VNIR-HSI system (400-1000 nm). The hyperparameters of CNN were fine-tuned using bayesian optimization (BO), which resulted in an 18.42% reduction in the prediction root mean squared error (RMSE) compared to the traditional PLSR model. Additionally, the SHAP method was applied to interpret the CNN model and assess the contribution of variable wavelengths. The CNN model based on important wavelengths was used to visualize spatial distribution of firmness in sweetpotato samples. Though HSI has emerged as a promising tool for many agricultural applications, the technology faces difficulty in being directly used in a real-time system due to the extensive time needed to process large volumes of data. Consequently, the development of a simple, compact, and cost-effective imaging system is not possible with the current HSI systems. Therefore, the overall goal of the third part of the study was to reconstruct hyperspectral images from RGB images through deep learning for agricultural applications. Specifically, this part of the study used hyperspectral convolutional neural network - dense (HSCNN-D) to reconstruct hyperspectral images from RGB images for predicting SSC in sweetpotatoes. The algorithm reconstructed the hyperspectral images from RGB images, with the resulting spectra closely matching the ground-truth. The PLSR model based on reconstructed spectra outperformed the model using the full spectral range, demonstrating its potential for SSC prediction in sweetpotatoes. In the final part of the study, three different hyperspectral reconstruction algorithms, such as HSCNN-D, hierarchical regression network (HRNET), and multi-Scale transformer plus plus (MST++), were compared to assess the DMC of sweetpotatoes. Among the tested reconstruction methods, HRNET demonstrated superior performance, achieving the lowest mean relative absolute error (MRAE) of 0.07, RMSE of 0.03, and the highest peak signal-to-noise ratio (PSNR) of 32.28 decibels (dB). Some key features were selected using the genetic algorithm (GA), and their importance was interpreted using XAI. PLSR models were developed using the RGB, reconstructed, and ground truth (GT) data. The visual and spectra quality of these reconstructed methods was compared with GT data, and prediction maps were generated
Optimizing rebuffering time under dynamic user behavior
No full textSubmission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2027-05-01The student, Jiayu Zhu, accepted the attached license on 2025-04-14 at 15:01.The student, Jiayu Zhu, submitted this Thesis for approval on 2025-04-14 at 15:15.This Thesis was approved for publication on 2025-04-16 at 10:38.DSpace SAF Submission Ingestion Package generated from Vireo submission #21713 on 2025-10-19 at 19:52:47Adaptive bitrate streaming (ABR) and quality of experience (QoE) metrics are proposed to enhance video streaming quality across various Internet connections. Traditional approaches to evaluating these metrics often ignore common user behaviors like seeking, jumping, or replaying video segments, leading to gaps in QoE understanding. Addressing this, we first collected thousands of audience retention curves from Bilibili, offering a thorough view of viewer engagement and diverse watching styles, to prove that the audience does not watch a video in full. Our analysis also reveals notable behavioral differences across video categories, with Bilibili showing trends of early video abandonment, possibly influenced by platform-specific factors and shorter video formats. This enhanced grasp of user engagement aids in refining ABR and QoE metrics. To address the QoE reduction due to the nature of dynamic use behavior, we thus propose StallFreeSeek streaming system, which utilizes the good network conditions given by increased deployment of fiber-to-the-home and 5G services, as CDN appliances inside of ISPs drive down round-trip time. The intuition behind StallFreeSeek (SFS) is to prefetch small chunks densely distributed across the video, allowing immediate playback on almost any skip, and exploit strong network performance to fetch ever-larger chunks before each previous chunk finishes playback. Our evaluations show that SFS improves Quality-of-Experience and stall times in suitable network conditions while wasting less buffered content, and never performs worse than dash.js across thousands of runs. Our evaluations show that across video genres, models of user seeks, and in real-world user studies, SFS is never inferior to dash.js in QoE, stall time, or buffer waste, and when network conditions allow, performs significantly better