BYU’s data-driven research map.

Work. Machine learning, knowledge representation, generative AI, human-centered AI, and the societal effects of language models.

Why it matters. The lab studies how AI systems can become more useful, interpretable, responsible, and compatible with human goals.

Work. Deep learning, probabilistic programming, reinforcement learning, large language models, and computational approaches to social behavior.

Why it matters. The work connects foundational ML with questions in social science, medicine, communication, and human behavior.

Work. Offline and risk-sensitive reinforcement learning, representation learning, causal inference, scientific discovery, healthcare, and LLM post-training.

Why it matters. The group targets reliable decisions in settings where experimentation may be costly, constrained, or unsafe.

Work. Human–robot interaction, bio-inspired robot swarms, multiagent systems, graph data science, and artificial intelligence.

Why it matters. The research helps people supervise, collaborate with, and understand autonomous systems.

Work. Machine translation for low-resource languages, natural-language processing, speech, and signed-language technologies.

Why it matters. The lab extends modern language technologies to communities that do not have the massive datasets available for major languages.

Work. NLP, large language models, big-data systems, and large government or social-media datasets for modeling public-health patterns.

Why it matters. The group turns heterogeneous, high-volume data into evidence that can inform public-health understanding and response.

Work. AI agents, cooperation, strategic interaction, contextual bandits, adversarial threats, and simulations of networked societies.

Why it matters. The work examines how intelligent agents cooperate, compete, and form stable arrangements with humans and with one another.

Work. Deep learning and computational creativity across machine-generated art, music, language, and other creative artifacts.

Why it matters. Creativity provides a demanding test of what an AI system can represent, learn, evaluate, and generate.

Work. Information retrieval, recommender systems, retrieval-augmented generation, embeddings, LLM evaluation, and mathematical question answering.

Why it matters. The lab improves how language models find, rank, connect, and reason over external information.

Work. Deep learning for automated transcription of historical handwriting and accessible family-history applications.

Why it matters. Automated transcription makes large collections of historical documents more searchable and useful for family-history research.

Work. Machine learning, data mining, reinforcement learning, and methods that help learned systems generalize to new data.

Why it matters. The research seeks more capable and efficient learning systems that transfer beyond a single benchmark or dataset.

Work. Computer vision, recognition, visual representations, and fine-grained classification.

Why it matters. The work helps AI distinguish visually subtle categories and turn imagery into structured information.

Work. Natural-language processing and machine learning for activities and devices in the physical world, with an emphasis on human interaction.

Why it matters. The research connects statistical AI with systems people directly use, interpret, and experience.

Work. Mathematical systems theory, predictive modeling, networks, feedback, and control.

Why it matters. The work supplies mathematical foundations for learning models from data and using them in decision and control systems.

Work. Human–AI interaction, neural-network interpretability, graph learning, higher-order interactions, generative AI, and diffusion models.

Why it matters. The group develops statistical approaches for understandable AI and complex structured data across multiple applications.

Work. Data-efficient ML, active learning, graph-based and semi-supervised learning, core-set selection, and statistical learning theory.

Why it matters. The research aims to reduce the labeled data and computation machine-learning systems need while retaining mathematical guarantees.

Work. Scientific ML, data assimilation, model reduction, uncertainty quantification, and data-driven surrogate models.

Why it matters. Surrogates and reduced models can make high-fidelity simulation practical for optimization, inverse problems, and digital twins.

Work. Mathematical and physical explanations of machine-learning behavior, including double descent and its implications for model design.

Why it matters. The collaboration connects generalization behavior to ideas from mathematics and physics that may guide better architectures and training.

Work. Bayesian spatial and spatiotemporal statistics, environmental modeling, uncertainty, Gaussian-process surrogates, and neural approaches to spatial problems.

Why it matters. The methods make environmental and geospatial predictions more scalable while representing uncertainty explicitly.

Work. Bayesian hierarchical and nonparametric modeling applied to sports, human performance, and health.

Why it matters. The work converts athletic performance data into interpretable evidence for tactics, training, and coaching decisions.

Work. Neural networks, transformers, computer vision, modeling, and simulation for identifying nanostructures in bacterial tomograms.

Why it matters. Automated image analysis can surface biological structure across collections that would be prohibitively labor-intensive to inspect by hand.

Work. Machine learning for acoustic-source detection, localization, ocean-environment inference, and estimating seafloor properties from sound.

Why it matters. The methods support passive sonar, ocean sensing, and robust inference from large sensor arrays in changing environments.

Work. Medical foundation-model evaluation, Bayesian panel predictions for prostate-cancer diagnosis, computational molecular methods, and protein engineering.

Why it matters. The group studies medical-AI robustness while applying computation to diagnosis, proteins, and therapeutics.

Work. Statistical inference and deep learning for astronomy, including faint moving-object detection in large image collections.

Why it matters. The work helps astronomers identify faint Solar System objects that conventional search methods may miss.

Work. Explainable machine learning for classifying human crowd reactions from sound, alongside broader acoustic-data modeling.

Why it matters. Objective crowd-state estimates may inform venue safety, event management, and research on collective human behavior.

Work. High-resolution computational analysis and visualization of neural recordings to characterize seizure initiation and progression.

Why it matters. The tools make subtle spatial and temporal patterns in dense neuroscience measurements easier to inspect.

Work. Physics-informed machine learning, hybrid models, dynamic optimization, predictive control, energy, drilling, nuclear systems, manufacturing, and digital twins.

Why it matters. PRISM combines first-principles models with data to support engineering design and real-time operation.

Work. Computer vision, pose estimation, biomechanics, video analysis, wearable and performance data, and AI-assisted movement measurements.

Why it matters. The project translates technical estimates into useful coach-facing information while training students in responsible data practices.

Work. Machine and deep learning, optimization, aerodynamics, aircraft design, wind-energy systems, and rapid flow-field prediction.

Why it matters. Learned models can augment expensive aerodynamic simulation in engineering design and optimization.

Work. Robotics and machine learning for weeding, harvesting, crop testing, and combining biomechanical experiments with ML to improve plant structure.

Why it matters. The research applies learning both to the machines that farm and to the plants being cultivated.

Work. Smart manufacturing, connected machine data, predictive analytics, automation, and digital twins.

Why it matters. The work helps factories detect, predict, and optimize machine and process behavior using real operational data.

Work. Geospatial data science, hydrologic modeling, ML post-processing, APIs, model-context protocols, and multi-agent LLM systems for water data.

Why it matters. The lab makes national-scale water information easier for researchers and decision-makers to query, interpret, and use.

Work. Supervised machine learning converts wearable skin-strain measurements into estimates of lumbar-spine motion.

Why it matters. The approach may enable lower-cost, noninvasive study of spinal movement in real-world chronic back-pain research.

Work. Artificial intelligence, robotic vision, real-time machine vision, automated visual inspection, and high-performance visual computing.

Why it matters. The research joins computer-vision algorithms with the hardware and real-time systems needed outside the laboratory.

Work. Computational chemistry and data-driven methods for reaction mechanisms and catalyst design, including ML predictions of catalyst behavior.

Why it matters. Computation can narrow the laboratory search space and reduce the experimental cost of discovering industrial catalysts.

Work. Machine learning for gene-expression data, cancer and biomedical prediction, reusable omics datasets, and AI-assisted search of public biological data.

Why it matters. The work makes large genomic resources easier to find, reuse, and convert into biomedical hypotheses.

Work. Transcriptomics, RNA sequencing, pathogen genomics, machine learning, biomarker discovery, and computational analysis of infection and chronic disease.

Why it matters. The lab identifies disease signatures and possible biomarkers or therapeutic targets within large biological datasets.

Work. Statistical genetics, AI and ML, phenotype extraction from electronic health records, and precision-medicine analysis.

Why it matters. The research links genetic and clinical data to more individualized estimates of risk and treatment response.

Work. Machine learning, biological collections, genomics, biodiversity, and conservation-oriented analysis.

Why it matters. Computational methods make patterns in rapidly expanding biological and genomic collections available for evolutionary and conservation research.

Work. A generative-AI communication system that listens to disrupted speech from people with aphasia and suggests possible well-formed utterances.

Why it matters. The pilot explores whether language models can support more natural conversation after stroke or neurological injury.

Work. Machine learning, computer vision, historical-record linkage, census data, and interconnected family-history datasets.

Why it matters. The work supports family history and creates longitudinal data for research on people and communities across generations.

Work. LLM-generated personas and synthetic populations for studying political behavior, public opinion, survey questions, bias, and human responses.

Why it matters. Synthetic simulations may help social scientists explore hypotheses and pretest questions before costly human-subject studies—while raising important validity questions.

Work. Automated essay grading and AI-based instructional support for feedback, questioning, and assessment.

Why it matters. The research examines how formative feedback might reach more students without a proportional increase in instructor workload.

Work. How AI, robotics, programming, 3D printing, simulations, and related technologies support science teaching and learning.

Why it matters. The group evaluates when technology improves scientific reasoning rather than merely adding a new tool to the classroom.

Work. The assumptions, limitations, and conceptual framing of artificial intelligence in educational research and scholarship.

Why it matters. The work asks educators to examine the educational problems and values beneath AI adoption, rather than treating AI as an automatic solution.

Work. Natural-language processing and machine learning for language-learning applications, with particular expertise in Russian and other languages.

Why it matters. Intelligent language tools can provide targeted practice, analysis, and feedback beyond conventional instructional software.

Work. Computational linguistics, machine learning, cognitive modeling, neurolinguistics, information extraction, and textual ontologies.

Why it matters. The research turns linguistic structure into computational representations for language understanding and knowledge extraction.

Work. How companies use data, experimentation, and AI for innovation and strategic decisions, including comparisons of AI recommendations and human judgment.

Why it matters. The research identifies when data and AI improve strategic decisions—and when they may impede exploration or innovation.

Work. Generative-AI adoption and resistance, chatbots, conversational agents, cybersecurity training, and effects of generative AI on learning.

Why it matters. The work provides evidence for designing conversational systems people can appropriately trust, use, and learn from.

Work. Generative-AI search behavior, resistance to AI, AI-mediated interviews, and wearable technologies.

Why it matters. The research helps organizations design AI-enabled search, hiring, and decision systems around actual human behavior.

Work. Retrieval-augmented research assistants, generative AI in auditing and accounting, governance, internal controls, analytics education, and AI skills.

Why it matters. The work helps accountants audit, govern, evaluate, and productively use generative systems.

Work. LLM assistants for conceptual modeling, characterizing research contributions, and studying how generative systems learn and unlearn concepts.

Why it matters. The work may make complex information-system modeling and knowledge representation more accessible and efficient.

Work. Keystroke dynamics, mouse movement, identity verification, fraud detection, behavioral prediction, and distinguishing human from AI-assisted authorship.

Why it matters. Passive behavioral signals may support cybersecurity, authentication, online research, and fraud detection without relying only on passwords.

Work. Machine learning and textual analysis of financial disclosures, sentiment, unstructured corporate information, and capital-market behavior.

Why it matters. The research extracts signals from documents that conventional accounting ratios and structured databases may not capture.

Work. Modern statistical and machine-learning methods for studying corporate disclosure and how investors retrieve and process information.

Why it matters. The work explains how the structure and availability of information shape market participants and investment decisions.