Digital Commons Network^™

Pre-Training And Interpretability In Deep Learning Models, Weizhi An

Computer Science and Engineering Dissertations

In the evolving field of artificial intelligence, the efficacy of deep learning models is often gated by the quality of their training and the clarity of their decision-making processes. This dissertation addresses these crucial challenges by focusing on two key areas: enhancing pre-training strategies and improving model interpretability. Our approach is twofold, integrating novel pre-training methodologies that embed domain-specific knowledge early in the model training process, and developing advanced techniques for disentangling and clarifying the decision-making mechanisms within these models. The first direction of our research employs MoDNA, a motif-oriented pre-training framework specifically designed for DNA language models. By leveraging …

Uncertainty Quantification, Propagation & Conjunction Assessment In Orbital Mechanics Using Generalized Polynomial Chaos Expansion & 2-Dimensional Conjunction Plane Analysis Techniques, Monalisa Karim

Mechanical and Aerospace Engineering Theses

Uncertainties, that are inherent to dynamic models, can be associated with state initial conditions, force modelling errors, navigation and actuation errors. In system modelling stochastic differential equations are used to represent dynamic phenomena with uncertainties, for which the solutions are probability density functions of quantities of interest characterizing the realization of the stochastic processes. In Polynomial Chaos Expansion (PCE) propagation, these solutions are represented as weighted sums of multivariate spectral polynomials that are functions of the input random variables. Generalized polynomial chaos expansion (gPC) is an extension to the original homogenous PCE which projects the random solution onto a basis …

The Transport Of New Make Spirits Through American White Oak, Taylor E. Scott, Jarrad Gollihue, Michael P. Sama, Michael Shaffer, Brad J. Berron

Biosystems and Agricultural Engineering Faculty Publications

Why was the work done: The reduction of maturation loss is a significant opportunity for increased revenue in aged spirits and varies through several factors. This work aims to better understand maturation loss by isolating and measuring spirit flow through wood.

How was the work done: Bourbon stave disks were fabricated by computer numeric control milling and loaded into a sealed module consisting of flanges holding the stave disk over a liquid filled pipe. The modules were filled with new make spirit at 50, 62.5 or 70% ABV, with either liquid or vapour contact with the stave disk. Further, the …

Consensus And Controversies Of International Guidelines For The Diagnosis, Surveillance, And Management Of Fetal Growth Restriction: An Updated Comparison, Daniele Diane Mascio, Suneet P. Chauhan, Tullio Ghi, Asma Khalil, Juliana G. Martins, Sara Sorrenti, Tamara Stampalija, Fabrizio Zullo, Francesc Figueras

Department of Obstetrics & Gynecology Faculty Publications

OBJECTIVE: To compare areas of consensus and disagreements across contemporary international and national guidelines on the diagnosis, surveillance, and management of fetal growth restriction (FGR).

DATA SOURCES: Electronic searches of MEDLINE from database inception up to March 2026 using MeSH terms and keywords related to FGR and guidelines. STUDY ELIGIBILITY CRITERIA: Critical, structured comparison of national or international guidelines on FGR published since 2010. Final inclusion required unanimous agreement from all authors.

STUDY APPRAISAL AND SYNTHESIS METHODS: Pre-specified extraction across domains: definition; prediction/prevention; surveillance tools and frequency; delivery timing and mode; and labor induction methods. Dual data …

Osmotic Dilators Vs. Foley Balloons: Bayesian Secondary Analyses Of The Dilafol Trial, Antonio F. Saad, Elenir B. C. Avritscher, Claudia Pedroza, Xiao Han, George R. Saade

Department of Obstetrics & Gynecology Faculty Publications

Background: Dilapan-S, a synthetic osmotic dilator used for pre-induction cervical ripening, has proven to be as effective as the Foley balloon in terms of vaginal versus cesarean delivery and has been associated with increased patient satisfaction. Nevertheless, its ability to lower cesarean rates and its cost-effectiveness remain poorly studied.

Objectives: To estimate the likelihood that synthetic osmotic dilators reduce cesarean deliveries and to evaluate their health system costs (2026 US$) and incremental cost-effectiveness compared to the Foley balloon during cervical ripening.

Study Design: Secondary Bayesian analyses were conducted using data from the single-center DILAFOL randomized controlled trial, which enrolled 419 …

Enhancing Explainable Ai For Medical Imaging: Improved Lime Interpretation With Influence Mapping, Abiha Tahsin Chowdhury, Dhanush Bavisetti, Daniel B. Hier, Rahul Dubey, Tayo Obafemi-Ajayi

Electrical and Computer Engineering Faculty Research & Creative Works

The integration of artificial intelligence (AI) into medical imaging is progressing rapidly. It is essential for these AI tools to be transparent, interpretable, and explainable to gain the trust of clinicians and regulators. Current state-of-the-art explainable AI (XAI) techniques in imaging includes Local Interpretable Model-Agnostic Explanations (LIME), Shapley Additive Explanations (SHAP), and Gradient-weighted Class Activation Mapping (Grad-CAM). Recent studies have shown that LIME often suffers from inconsistency and unreliability which limits their utility in sensitive fields like medical imaging. This paper proposes Influence Map based Explanation (IME), an enhanced variant of the original LIME framework, that aggregates multiple runs to …

Event-Based Predefined-Time Synchronization For Complex Networks With Deception Attacks: An Asynchronously Intermittent Strategy, Lei Xue, Jiong Yu, Haoyu Zhou, Yongbao Wu, Jian Liu, Donald C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This article studies the practical predefined-time synchronization (PPTS) for complex networks (CNs) under deception attacks based on the asynchronously intermittent event-triggered control (AIE-TC). Notably, AIE-TC effectively integrates the advantages of asynchronously intermittent control (AIC) and event-triggered control, where AIC provides each subsystem node with independent control and rest intervals. Besides, all synchronization errors of the CNs converge to an adjustable neighborhood within the predefined time by designing a bounded time-varying function into the controller. Moreover, this article considers that the transmission network is subjected to stochastic deception attacks modeled by a Markov process, which captures the state-driven dynamic transition characteristics …

Femtosecond-Laser-Inscribed Cladding Fiber Bragg Gratings: Architectures, Design Trade-Offs And Future Sensing Opportunities, Koustav Dey, Rex E. Gerald, Chen Zhu, Jie Huang

Electrical and Computer Engineering Faculty Research & Creative Works

Femtosecond (fs) laser inscription enables fiber Bragg gratings (FBGs) to be positioned beyond the fiber core, including within the cladding, thereby expanding control over modal interactions and directional sensing. These cladding fiber Bragg gratings (CLFBGs) extend FBG functionality through interactions with core, cladding, and evanescent fields. Their operation does not involve fundamentally different grating physics; rather, core-mode, cladding-mode, and coupled-mode responses depend on power distribution, modal overlap with the laser-inscribed refractive-index modulation (RIM), grating position, fiber geometry, modal excitation, and interrogation configuration. This Perspective reviews advances in CLFBG fabrication, optical characteristics, sensing mechanisms, and applications. Cladding-waveguide-assisted, evanescent-field-coupled, and direct cladding-mode-excitation …

Temperature Compensation In Loop And Patch Fss Strain Sensors: Analysis And Experimental Validation, Swathi Muthyala Ramesh, Kristen M. Donnell

Electrical and Computer Engineering Faculty Research & Creative Works

Frequency selective surfaces (FSSs) are arrays of conductive elements or apertures that exhibit frequency-dependent reflection and transmission properties. Their electromagnetic response is influenced by geometry and environmental conditions, making them attractive for wireless strain-sensing applications. However, temperature variations can produce frequency shifts similar to those caused by strain, reducing measurement accuracy. This work investigates the effects of intrinsic temperature compensation on two common FSS unit cell geometries—loop and patch—through comprehensive simulation analysis. The results show that loop-based cells offer superior thermal stability, while patch-based cells provide greater strain sensitivity, illustrating the trade-off between thermal robustness and mechanical responsiveness. A patch-type …

Large Language Models For Neurology: A Mini Review, Donald C. Wunsch, Daniel B. Hier

Electrical and Computer Engineering Faculty Research & Creative Works

Large language models have the potential to transform neurology by augmenting diagnostic reasoning, streamlining documentation, and improving workflow efficiency. This Mini Review surveys emerging applications of large language models in Alzheimer's disease, Parkinson's disease, multiple sclerosis, and epilepsy, with emphasis on ambient documentation, multimodal data integration, and clinical decision support. Key barriers to adoption include bias, privacy, reliability, and regulatory alignment. Looking ahead, neurology-focused language models may develop greater fluency in biomedical ontologies and FHIR standards, improving data interoperability and supporting more seamless collaboration between clinicians and AI systems. Two future developments have the potential to be particularly impactful: (1) …

Safety Aware Continual Reinforcement Learning-Based Output Tracking Control Of Nonlinear Continuous-Time Systems, Irfan Ganie, Sarangapani Jagannathan

Electrical and Computer Engineering Faculty Research & Creative Works

An output feedback (OF)-based control scheme utilizing both a scalable multilayer neural network (MNN) observer and actor–critic MNN via integral reinforcement learning (IRL)/adaptive dynamics programming (ADP) approach for a class of nonlinear systems with output constraints is introduced. The proposed observer, critic, and actor MNN weight updates are derived using a singular value decomposition (SVD) of MNN activation function gradient along with output error, Bellman and control input errors, respectively. Next, the approach incorporates continual learning (CL), utilizing a penalty function in the weight update laws for both actor–critic MNNs to consolidate knowledge from previous tasks and enhance learning in …

High-Resolution, Fast-Response Optical Fiber Temperature Sensor With A Large Measurement Range Based On Fiber-Tip Alumina Fabry-Pérot Interferometer, Ruimin Jie, Chen Zhu, Robert Abbott, Michael Davis, Xiong Zhang, Jie Huang

Electrical and Computer Engineering Faculty Research & Creative Works

We present an alumina-tip optical fiber Fabry-Pérot interferometric temperature sensor exhibiting high-temperature performance, rapid response, and high resolution. Fabricated by fusion splicing an alumina micro disk directly to a single-mode fiber, the sensor achieves robust, stable operation without complex fabrication processes or adhesives. Experimental evaluation confirms a measurement range extending to 1000°C, with sensitivity of 28.66 pm/°C, a resolution of 0.042°C, and a rapid response time of approximately 13 ms. Compared to state-of-the-art optical fiber FPI sensors, our alumina-tip sensor offers superior overall performance, effectively addressing critical demands for high-resolution, fast-response temperature measurement in extreme environments including aerospace, structural monitoring, …

Safe Optimal Control Framework For Cooperative Manipulation Of Objects In Human–Robot Teams, Irfan Ganie, Sarangapani Jagannathan

Electrical and Computer Engineering Faculty Research & Creative Works

This article introduces a distributed deep neural network (NN)-based adaptive control framework for cooperative object manipulation in human–robot teams with unknown agent dynamics by using three distinct multilayer NN observers (MNNOs). The first observer, termed the reference point estimator, enables each robotic agent to estimate the object's reference center using consensus-based learning, even without direct access to global reference trajectories. The second observer, referred to as the human force-to-trajectory estimator, uses human-applied forces to infer the intended position, velocity, and acceleration of the object, enabling real-time estimation of human intent. Together, these two observers allow distributed estimation of human-intended motion. …

Distributed Temperature Sensing In The Spray-Cooled Shell Of A 150-Ton Dc Electric Arc Furnace Using Brillouin Optical Fiber Technology, Farhan Mumtaz, Yeshwanth Reddy Mekala, Koustav Dey, Rony Kumer Saha, Ogbole Collins Inalegwu, Manoj Kumar Pullagura, Bohong Zhang, Muhammad Roman, Nicholas Dionise, Zane Voss, Jeffrey D. Smith, Ronald J. O'Malley, Rex E. Gerald, Jie Huang

Electrical and Computer Engineering Faculty Research & Creative Works

This paper presents the deployment and validation of a Brillouin - distributed temperature sensing (DTS) system for real-time thermal monitoring of the spray-cooled upper shell of a 150-ton direct current Electric Arc Furnace (DC EAF) at Big River Steel Plant, Osceola, AR, USA. A four-channel Brillouin DTS system from OZ Optics was employed, with one active channel instrumented using an in-house-fabricated Brillouin scattering-depressed single-mode optical fiber (SMF28e+). The 60 m optical fiber sensor was fabricated, with 20 m allocated for thermal measurement and 40 m used as lead-in fiber to isolate the interrogator from the furnace environment. The fiber was …

Self-Calibrating Uav Navigation: Reinforcement Learning Approaches For Horizontal Trajectory Estimation, Shirin Nasr-Esfahani, S. Jagannathan

Electrical and Computer Engineering Faculty Research & Creative Works

Accurate unmanned aerial vehicle (UAV) trajectory estimation is essential for autonomous navigation, particularly in GPS-denied environments. Visualodometry and simultaneous localization and mapping (SLAM) approaches require precise camera intrinsic parameters, which are typically obtained through predefined or offline calibration. Instead, in this work, we propose a reinforcement learning (RL)-based self-calibration framework that estimates camera intrinsic parameters directly from monocular video sequences, without requiring prior knowledge of the camera, environment, or calibration targets. This intrinsic parameter estimation is then leveraged to achieve robust UAV trajectory estimation using only video data. We formulate the problem as a sequential decision-making task, where an RL …

Fixed-Time Consensus Tracking For Nonlinear Multi-Agent Systems Under Aperiodically Intermittent Control, Lei Xue, Tong Wu, Wenwen Jia, Donald C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This article explores the problem of fixed-time consensus tracking (FT-CT) for nonlinear multi-agent systems utilizing the a periodically intermittent control (AIC) strategy. In contrast to existing control algorithms, the proposed algorithm utilizes the AIC strategy instead of the conventional continuous-time control strategy, effectively reducing the consumption of communication resources. Moreover, the problem of intermittent FT-CT is well handled by proposing the average control rate of the AIC strategy. Two theorems based on the cases of directed and undirected graphs are proposed, respectively. Finally, the validity of these results is confirmed through numerical simulations on a general nonlinear system and a …

Rf-Attennet: A Hybrid Attention-Enhanced Network For Mixed Signal Classification In Uav Swarm Detection, Prajoy Podder, Mohammad Atikur Rahman, Maciej Zawodniok, Sanjay Madria

Electrical and Computer Engineering Faculty Research & Creative Works

The continuous increase of UAVs, particularly in swarms, creates significant challenges for security and airspace regulation. Traditional RF fingerprinting methods struggle to detect and classify UAV swarms due to overlapping signals and interference. This study introduces RF-AttenNet, a hybrid deep learning model designed to classify mixed UAV signals by analyzing composite RF spectrograms. RF-AttenNet uses dual attention mechanisms, channel and spatial attention to focus on critical spectral features, enabling the model to effectively separate and identify overlapping UAV signals. We have developed custom composite UAV datasets that simulate real-world swarm interference, incorporating both single and mixed UAV classes. RF-AttenNet achieves …

Sapphire Optical Fiber Bragg Grating Sensors Based On Dispersive Microwave-Photonic Frequency-Time Domain Analysis, Ruimin Jie, Chen Zhu, Farhan Mumtaz, Koustav Dey, Bohong Zhang, Jie Huang

Electrical and Computer Engineering Faculty Research & Creative Works

Sapphire fiber Bragg gratings (SFBGs) have attracted growing interest for high temperature sensing in harsh environments, yet their interrogation typically relies on optical spectrum measurements, demanding a high-resolution optical spectrum analyzer (OSA) that is bulky, expensive, and constrained in acquisition speed. Moreover, the inherently multimode nature of sapphire fiber further complicates spectrum-based demodulation, thereby limiting the achievable sensing resolution. In this paper, we propose and experimentally demonstrate a microwave-photonic interrogation approach for SFBG sensors. Instead of measuring the optical reflection spectrum, the complex frequency response in the microwave domain of an SFBG is acquired using a vector network analyzer (VNA) …

Multi-Agent Reinforcement Learning Driven Package Pdn Design Automation, Haran Manoharan, Chulsoon Hwang

Electrical and Computer Engineering Faculty Research & Creative Works

The design of package-level power delivery networks (PDNs) has become increasingly challenging as modern high-performance systems demand higher currents. Existing PDN design flows treat ball map assignment, stackup selection, and power plane routing as separate, largely manual steps, leading to long iteration times and limited scalability. This work proposes a unified and automated package PDN design framework based on multi-agent reinforcement learning (MARL). Each power domain is modeled as an agent, with specialized agents responsible for ball map assignment, routing layer selection, and power plane synthesis. A central controller coordinates agent decisions using a global reward that captures electrical and …

Analytical And Semi-Analytical Modeling Of Solar Cells Using The Lambert W Function: A Comprehensive Review Of Equivalent Circuits, Adel El-Shahat, Martin Ćalasan, Snežana Vujoševic, Shady H. E. Abdel Aleem

Engineering Technology Faculty Publications

The modeling of photovoltaic (PV) cells through equivalent circuits forms a central element in the analysis, simulation, and optimization of solar energy systems. Traditional approaches often depend on iterative numerical methods to solve the implicit current–voltage (I–V) equations. In contrast, the Lambert W function has emerged as an effective mathematical tool that enables closed-form or semi-analytical expressions for a wide range of PV models. This paper presents a Lambert W-centered review of analytical and semi-analytical formulations for PV equivalent-circuit models, covering classical single-diode and multi-diode structures and modern variants incorporating additional elements, voltage-dependent parameters, and topology rearrangements. The models are …

Interpretable Machine Learning For Bridge-Pier Scour Prediction And Flood Resilience, Adil Khan, Dalya Ismael

Engineering Technology Faculty Publications

Bridge-pier scour is a leading cause of flood-induced bridge failure, yet practice still lacks transparent, physics-informed tools that link data-driven prediction with design guidance. This study develops an interpretable, physics-aware machine-learning framework to predict equilibrium scour depth and translate those predictions into actionable strategies for flood-resilient infrastructure. Using the 2014 U.S. Geological Survey Pier-Scour Database (569 laboratory cases), five models: Gradient Boosting, AdaBoost (Tree), XGBoost, Gaussian Process (RBF kernel), and Kernel Ridge (polynomial), were trained and evaluated with K-fold cross-validation. Model performance was evaluated using R², RMSE, and MAE. Gradient Boosting performed best, achieving training and testing R² of 0.99 …

Generative Ai And Llm Applications In Renewable Energy And Smart Grids: A Systematic Review For The Sustainable Energy Transition, Umit Cali, Ugur Halden, Merlinda Andoni, Ferhat Ozgur Catak, Si Chen, Benoit Couraud, Emre Kantar, Samuel Knapper, Ibrahim Kucukdemiral, Huseyin Kusetogullari, Murat Kuzlu, Yashar Mousavi, Sonam Norbu, Taha Selim Ustun, David Flynn

Engineering Technology Faculty Publications

The global energy transition toward decarbonization and digitalization requires advanced methods to manage decentralized, data-intensive cyber-physical energy systems. This systematic review analyzes 106 research studies on Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) in renewable energy and smart grids, organized into seven application clusters covering forecasting, system design, operation, reliability, data and cybersecurity, and energy markets. The review situates these applications within a Cyber-Physical-Social Systems (CPSS) framework. Results show that GANs dominate current applications (47.2%), followed by LLMs (10.4%) and VAEs (9.4%), with growing adoption of diffusion and score-based models (7.5% each). Selected studies report improved probabilistic forecasting …

Explainable Ai-Driven Predictive Maintenance Curriculum For Smart Manufacturing, Abdullah Al Mamun, Murat Kuzlu, Katherine Smith, Vukica Jovanovic, Dalya Ismael, Adel El-Shahat, Angela Sicaja, Md. Hedayetul Islam Chy

Engineering Technology Faculty Publications

Nowadays, Industry 4.0 has transformed manufacturing industries into a data-rich system driven by IoT, automation, and Artificial Intelligence (AI). Within this context, Predictive Maintenance (PdM) provides a proactive strategy that leverages heterogeneous sensor data such as vibration, acoustic, electrical, and visual signals along with historical performance and advanced analytics to forecast equipment failures before they occur. Usually, AI-driven PdM (AI-PdM) enhances this capability by integrating AI-based sensor analytics to automate fault prediction and optimize system reliability. However, traditional AI-PdM often functions as a “black box,” providing limited interpretability of its decision-making process and posing challenges for trust, validation, and human …

Wip: Arduino-Based Pbl To Foster Entrepreneurial Mindset, Nathan Q. Holland, Vukica M. Jovanovic

Engineering Technology Faculty Publications

This work-in-progress paper examines a semester-long project-based learning initiative begun in an introductory engineering course during the Fall 2025 semester. The project aims to enhance students' hands-on experience by integrating information literacy, the engineering design process (EDP), and an entrepreneurial mindset (EM) [1]. The objective is to boost student confidence in teamwork, technical problem-solving, and application of skills, preparing them for future courses in their discipline. Students in this introduction to engineering course received an Arduino Uno R3 Controller board kit and additional sensors. They were tasked to develop a device addressing engineering in the medicine challenge of their choosing. …

Improving Medical Diagnostics With Vision-Language Models: Convex Hull-Based Uncertainty Analysis, Ferhat Ozgur Catak, Murat Kuzlu, Taylor Patrick, Michel Audette

Engineering Technology Faculty Publications

In recent years, vision-language models (VLMs) have been applied to various fields, including healthcare, education, finance, and manufacturing, with remarkable performance. However, concerns remain regarding VLMs' consistency and uncertainty, particularly in critical applications such as healthcare, which demand a high level of trust and reliability. This paper proposes a novel approach to evaluate uncertainty in VLMs' responses using a convex hull approach on a healthcare application for visual question answering (VQA). For any VLM, temperature refers to a sampling parameter used in probabilistic generation, which controls the randomness of the model's output. The LLM-CXR model is selected as the medical …

Application Paths Of Semantic Modeling In Financial Fraud Detection And Risk Identification, Victor P. Gauthier, Daniel S. Wu

Computer Science Faculty Publications

Financial fraud and risk pose significant threats to economic stability and individual well-being. Traditional detection methods often struggle to keep pace with increasingly sophisticated fraudulent schemes. Semantic modeling, which focuses on understanding the meaning and relationships within data, offers a promising avenue for enhancing fraud detection and risk identification. This review paper explores the application paths of semantic modeling in this domain. We begin with a historical overview of fraud detection techniques, highlighting the limitations of traditional approaches. Subsequently, we delve into core themes, including knowledge graph-based fraud detection and semantic rule-based inference for risk assessment. We then compare and …

Towards Supporting Real-Time Estimation Of Vehicle Fuel Consumption And Co2 Emissions In Smart City Applications, Abrar Alali, Stephan Olariu

Computer Science Faculty Publications

This paper evaluates a simplified physics-based energy demand model designed to estimate vehicle fuel consumption and CO₂ emissions—a critical tool for sustainable transportation planning and smart city applications. Unlike data-driven regression models that lack generalizability for user-defined conditions or complex physics-based approaches that rely on extensive, often proprietary data, the simplified model is distinguished by its minimal parameter requirements, depending primarily on a single, overarching powertrain efficiency value. A key contribution is the comprehensive empirical evaluation of the simplified model against official Environmental Protection Agency (EPA) test data across multiple driving cycles and vehicle types, providing a rigorous validation previously …

Privacy-Preserving Federated Learning With Optimized Ensemble Weighting And Knowledge Distillation For Covid-19 Detection From Non-Iid Medical Imaging Data, Richard Annan, Hong Qin, Robert Newman, Madhuri Siddula, Letu Qingge

Computer Science Faculty Publications

Medical imaging enables rapid and accurate diagnosis of COVID-19, with CT scans proving especially effective. However, data privacy concerns limit collaborative model development across hospitals. To address this issue, we introduce a novel federated learning framework. It is referred to as Independent Knowledge Distillation with post-Ensemble Federated Learning (IKDEFL). Differential Privacy (DP) is integrated into the framework to improve privacy guarantees. Three DP mechanisms are evaluated. These include Fixed Gaussian, Gaussian Adaptive, and Tree Adaptive. The evaluation has been conducted on heterogeneous and Non-Independent and Identically Distributed (Non-IID) datasets. These datasets reflect real-world hospital scenarios. Results show that IKDEFL significantly …

Biosecure-Llm Framework: Protecting Llms From Cyberbiosecurity Threats And The Case For Independent Ai Safety Governance, Xavier-Lewis Palmer, Lucas Potter, Srdjan Lesaja, Sotirios Karathanasis, Mohammad Ghasemigol

Computer Science Faculty Publications

Large Language Models (LLMs) are becoming critical infrastructure in scientific, healthcare, and governmental contexts. As frontier AI laboratories increasingly partner with government agencies, a fundamental question arises: Who should control the safety and policy-enforcement layers that constrain model behavior? Current safety mechanisms (LLM guardrails) are typically designed for generic "harmlessness" and operate by detecting semantic patterns and refusing requests. However, they are inadequate governance instruments because they cannot implement auditable, domain-specific controls tied to external regulatory policy objects (e.g., control lists or rules governing personally identifying information). Even a perfectly aligned model is not able to express institution-specific policy without …

Isotropic Shrinkage Of Patterned Vacancies Enables Three-Dimensional Nanoprecise Metastructures For Visible Light Applications, Quansan Yang, Gaojie Yang, Takahiro Nambara, Hiroyuki Kusaka, Yuichiro Kunai, Alex C. Matlock, Corban Swain, Brett Pryor, Yannick Salamin, Daniel Oran, Hasindu Kariyawasam, Ramith Hettiarachchi, Dushan Wadduwage, Marin Soljačić, Peter T. C. Soflaei, Edward S. Boyden

Computer Science Faculty Publications

Three-dimensional metastructures with nanoscale feature sizes exhibit unique properties compared with structures with larger feature sizes, but are difficult to fabricate. Here we introduce implosion carving (ImpCarv), a method for photopatterning vacancies of complex geometry throughout materials, followed by isotropic shrinkage (>10-fold). ImpCarv works by photoactivating sensitizers to generate reactive oxygen species that cleave a swollen hydrogel at defined points, followed by controlled shrinkage via dehydration. ImpCarv creates three-dimensional metastructures where the refractive index of each point throughout a material can be specified with nanoscale precision via material presence or absence. By leveraging refractive index programmability for precise phase …