33760 research outputs found
Sort by
Essays on Human Capital and Labor Market Strategies: Navigating Organizational Growth, Social Issues, and External Pressures
No full textThis dissertation, comprised of three self-contained chapters, focuses on how internal and external environments impact retention, attraction, and allocation of human capital. It reveals how firms adapt their strategies on talent acquisition, capability development, and labor market positioning in response to evolving conditions. By focusing on the regulatory environment, social norms, and stakeholders, this dissertation contributes to a deeper understanding of mechanisms shaping mobility, acquisition and allocation of talent in dynamic labor markets. Chapter I examines how abortion-restrictive laws impact retention and relocation of high-skilled talent. The results from the unique dataset with 0.54 million scientists demonstrate that states with abortion-restrictive laws faced more difficulty attracting and retaining high-quality scientific talent and experienced a downward trend in their long-run scientific and innovative performance. The study illustrates how the regulatory environment can shape the relocation of human capital and regional competitiveness in science and innovation.
Chapter II investigates how firms adjust their interaction with labor markets and their diversity, equity, and inclusion (DEI) messaging around significant organizational events, such as an IPO. Text analysis of online job postings reveals that firms facing growth pressures post-IPO adopt inclusive language to compete for talent. The findings indicate that these trends were driven by shifts in firms’ strategic intent concerning human capital, given the changes in their resources and new growth objectives post-IPO. The study suggests that language around DEI in job postings is not merely a function of evolving social norms but is also a strategic response to labor market pressures.
Chapter III explores how young firms grow and shape their human capital capabilities by examining the influence of founders, investors, product and capital market dynamics on the capability growth strategies of firms. The findings reveal that investors were the dominant decision makers during the development of organizational capabilities, and founders’ impact weakened after their involvement. The study demonstrates that investor objectives and experiences can shape firms’ human capital and organizational growth strategies, potentially influencing exit outcomes.
Overall, this dissertation demonstrates how organizations adapt human capital strategies to navigate organizational growth, social issues, and external pressures. These insights contribute to ongoing discussions on how firms sustain competitive advantage in increasingly competitive labor markets while also highlighting broader economic and innovation consequences of workforce mobility and organizational strategies
Identification and Characterization of Viruses Associated with the Atlantic Blue Crab, Callinectes sapidus, Across Its United States Geographic Range
No full textAcross two hemispheres, including in the United States, the Atlantic blue crab, Callinectes sapidus, supports a multi-million dollar fishery and plays critical roles in benthic food webs. Recent blue crab population declines in the Mid-Atlantic highlight the need to better understand potential threats to the fishery, including infectious diseases. Viruses can cause disease outbreaks and mortality yet remain relatively understudied compared to other microbial pathogens of marine shellfish. Movement of harvested blue crabs between states may pose a threat of unknown magnitude to the health of wild crabs in the receiving location.
To gain a preliminary understanding of the viral communities associated with the blue crab, we performed a virus-enrichment protocol on specimens collected from six U.S. states, from New York to Texas along the U.S. Atlantic and Gulf Coasts. Enrichment was followed by RNA extraction, cDNA synthesis, and high-throughput sequencing. Bioinformatic tools were used to identify and characterize putative virus genomes in gill collections.
We discovered a total of 97 putative viral sequences, of which 29 were high-quality genomes with a k-mer coverage >10. Positive-sense RNA viruses were especially prominent, with provisional Picornavirales members constituting 19 of the 29 genomes. Beyond the Picornavirales, seven other positive-sense, two negative-sense, and one double-stranded RNA virus were identified. Finally, this survey yielded potential insights into viral geographic distributions.
This research expands our current knowledge of viruses associated with the blue crab across multiple U.S. states. Similar to several prior studies of crustacean viruses, this study identified a large number of entities that putatively belong to viruses of the Picornavirales. Future work will investigate whether the observation of regionally-restricted virus genomes is confirmed when additional crabs are screened using specific PCR assays. The knowledge generated here will enable investigation of the pathogenicity of novel blue crab viruses, which is especially relevant in the face of climate change and the interstate transport of marine resources. Ultimately, such research can help monitor and mitigate the spread of shellfish pathogens
Translating Natural Language to Visually Grounded Verifiable Plans
No full textTo be useful in household environments, robots may need to understand natural language in order to parse and execute verbal commands from novice users. This is a challenging problem that requires mapping linguistic constituents to physical entities and at the same time orchestrating an action plan that utilizes these entities to complete a task. Planning problems that previously relied on querying manually crafted knowledge bases can now leverage Large Language Models (LLMs) as a source of commonsense reasoning to map high-level instructions to action plans. However, the produced plans often suffer from model hallucinations, ignore action preconditions, or omit essential intermediate actions under the assumption that users can infer them from context and prior experience. In this thesis, we present our work on translating natural language instructions to visually grounded verifiable plans.
First, we motivate the use of classical concepts such as Linear Temporal Logic (LTL) to verify LLM-generated action plans. Our key insight is that combining a source of cooking domain knowledge with a formalism that captures the temporal richness of cooking recipes could enable the extraction of unambiguous, robot-executable plans. Building on this insight, we present Cook2LTL, a system that receives a cooking recipe in natural language form, reduces high-level cooking actions to robot-executable primitive actions through the use of LLMs, and produces unambiguous task specifications written in the form of Linear Temporal Logic (LTL) formulae. By expressing action plans in a formal language notation that adheres to a set of rules and specifications, we can generate discrete robot controllers with provable performance guarantees.
Second, we focus on grounding linguistic instructions to visual sensory information and we find that Vision Language Models (VLMs) often struggle with identifying non-visual attributes. Our key insight is that non-visual attribute detection can be effectively achieved by active perception guided by visual reasoning. To this end, we present a Perception-Action API that consists of perceptual and motoric functions. When prompted with this API and a natural language query, an LLM generates a program to actively identify attributes given an input image.
Third, we present NL2PDDL2Prog, a system that incorporates the Planning Domain Definition Language (PDDL) as an action representation as a means to combine the ability of LLMs to decompose a high-level task to a set of actions with the correctness of symbolic planning. Prior work has often relied on manually crafting PDDL domains, which can be a difficult and tedious process, especially for non-experts. To circumvent that, we obtain visual observations before and after the execution of an admissible action in our environment. We pass them to a VLM to derive the action semantics which are then sent to an LLM to infer the entire domain. Given the generated domain and an initial visual observation of the scene, the LLM can produce a PDDL problem description that is then solved by a symbolic planner and parsed into an executable python program. By binding the perceptual functions to action preconditions and effects explicitly modeled in the PDDL domain, we visually validate successful action execution at runtime, producing visually grounded verifiable action plans.
To demonstrate the applicability of our work in the real world, we design a ROS-powered robotic system capable of receiving natural language instructions and implementing simple cooking recipes on a kitchen counter. We begin by bootstrapping a proof-of-concept system where each object has an ArUco marker on it to facilitate tracking. At runtime, our system receives a natural language instruction, calls Cook2LTL or NL2PDDL2Prog and passes the produced action plan to a python Pick-and-Place API that we developed for recipe execution on a Sawyer robot. We include demonstrations of experiments we conducted on the simple recipe of making a burger using artificial food items in the laboratory.
To conclude, we discuss ongoing and future work on improving our existing systems. We plan to incorporate object affordances in the safeguarding formalisms we have used to verify LLM plans. This can be achieved by introducing a more fine-grained action representation to support lower-level primitive actions and produce affordance-aware policies. We also focus on supporting contact-rich manipulation tasks such as grasping delicate and deformable items that are not only ubiquitous in the kitchen but in other domains, too. By leveraging visual context, textual descriptions, and feedback from tactile sensors, we could learn a mapping from the visual and textual space to the amount of current required for compliantly grasping delicate objects. Finally, we are working on extending the tracking functionality of our robotic system by incorporating Deep Object Pose Estimation (DOPE) to track objects of known 3D models from the YCB dataset, without the need of markers
Creative Placemaking and Community Safety: Addressing Crime in Riverdale Park
No full textFinal report for PLCY400: Senior Capstone (Spring 2025). University of Maryland, College ParkThis project explores how creative placemaking can serve as a strategy to reduce crime rates and foster community connection in Riverdale Park, Maryland. Despite a general decline in crime across Prince George’s County, the public remains concerned about crime in Riverdale Park, particularly youth involvement in carjackings and violence. The town center, though rich in potential and local assets like small businesses and a farmers market, lacks inclusive, green, and welcoming public space. Our research focuses on how creative placemaking can activate this space to promote safety and belonging.
The resulting recommendation is for a recurring community arts event that offers youth safe, creative alternatives to crime. Using a mixed-methods approach, we conducted a literature review, case study analysis, and interviews with stakeholders, including local business owners, law enforcement, urban planners, and arts council representatives. Our findings suggest that community-centered design, cultural engagement, and environmental enhancements can reduce crime and support youth development.
While our study is limited by its short timeline, lack of long-term impact data, and gaps in community input, the recommendations are grounded in local voices and evidence-based practices. By activating Riverdale Park’s underutilized spaces, this project aims to foster social cohesion, elevate community identity, and lay the groundwork for sustainable, arts-driven revitalization.Prince George's County, M
Transgene-free Genome Editing in Poplar
No full textAchieving precise, transgene-free genome editing in perennial tree species remains a major challenge due to limitations in transformation efficiency, delivery methods, and regulatory concerns. In this study, we developed and evaluated genome editing strategies to optimize transgene-free genome editing in Populus tremula × alba. We compared the performance of two CRISPR-Cas9 base editor vector constructs—one standard (pLR5478) and one tagged with a mobile tRNA-like sequence (TLS; pLR5479)—for editing two key target genes: ALS (acetolactate synthase) and 4CL1 (4-coumarate-CoA ligase 1). Both vectors successfully generated edited plants; however, pLR5478 yielded a higher proportion of transgene-free edited lines compared to pLR5479. Notably, transgene-free plants edited at the 4CL1 locus using the TLS-tagged vector exhibited moderate-to-high editing efficiencies (40.9%–69.97%), outperforming those generated using the standard vector. These findings suggest a trade-off between editing efficiency and transgene-free recovery depending on vector architecture.We further explored biolistic delivery methods using deoxyribonucleic acid (DNA) and ribonucleoprotein (RNP) complexes to the axillary meristem. Comparisons between a low-pressure (650 psi) barrel-based system and a high-pressure (1350 psi) traditional rupture disc method revealed that the barrel system increased transformation efficiency with green fluorescent protein expression. Also, with RNP delivery, more plants were edited using the low-pressure barrel method, but plant samples bombarded using the traditional method had higher indel frequency. These results underscore the importance of optimizing physical delivery parameters to balance cell viability and editing efficacy.
In summary, this work demonstrates that TLS-enhanced vectors can improve base editing efficiency at certain loci in poplar but may hinder transgene-free plant recovery. Meanwhile, delivery of RNPs to meristematic tissues, though currently limited in efficiency, offers a viable path toward transgene-free genome editing. Thus, these results provide important insights into the development of precise, DNA-free genome editing systems in tree species and highlight practical trade-offs in vector design and reagent delivery that must be optimized for future deployment in forestry and bioenergy applications
Next-Gen AI: Advancing Watermarking, Algorithm Synthesis and Diverse Generative Strategies
No full textThis doctoral thesis explores critical advancements in generative artificial intelligence (AI) across three domains: enhancing security through watermarking, advancing algorithm synthesis, and pioneering diverse strategies for generative image tasks. The research contributes novel methodologies and unified frameworks that push the boundaries of AI capabilities.
The first part of the thesis addresses the growing need for AI security by introducing certified watermarking techniques to safeguard deep neural networks (DNNs) from intellectual property violations. These methods establish robust protections for model ownership and integrity, ensuring the secure deployment of AI innovations.
The second part investigates algorithm synthesis, expanding the computational and reasoning capabilities of neural networks. By enabling complex problem-solving and demonstrating advanced adaptability, this work redefines the role of neural networks as tools for sophisticated algorithmic design, transcending traditional applications in pattern recognition.
The third part focuses on diverse generative strategies for image creation and editing. It begins with Cold Diffusion, employing deterministic transformations to expand the operational mechanics of diffusion models. The research further enhances the generative process by enabling the creation of highly specific and contextually relevant imagery with minimal retraining, improving the flexibility and practicality of diffusion-based approaches. Finally, the thesis presents a unified framework for image reasoning and generation, leveraging next-token prediction with a vision encoder that produces discrete, non-lossy image embeddings aligned with language. This innovation enables a transformer-based architecture to support both high-precision image editing and advanced reasoning, paving the way for a cohesive and versatile AI design.
By addressing security, algorithmic adaptability, and generative innovation, this thesis contributes to the development of next-generation AI systems. It establishes a strong foundation for future advancements in AI technologies, ensuring secure, adaptable, and creative solutions for a wide range of applications
MODELING THE PAST: EMPLOYING CLOSE-RANGE PHOTOGRAMMETRY AND 3D MODELING METHODOLOGIES FOR DIGITAL HERITAGE PRESERVATION AT THE BRONZE AGE SITE OF BÉKÉS-VÁRDOMB IN TARHOS, HUNGARY
No full textThis paper aims to test and illustrate the effectiveness of close-range photogrammetry/3D modeling for documenting archaeological fieldwork and recovered artifacts during the Körös Consortium's “Understanding the Emergence of Cities” project at the Bronze Age site of Békés-Várdomb in Tarhos, Hungary. This thesis will detail the specific methods utilized to produce 3D models during the field project, document variables, and gauge the quality of models produced, to assess this method of documentation for public outreach, future archaeological research, and as a form of digital heritage preservation. The main research questions for this thesis include: “Why is photogrammetry not used more often within the field of archaeology”, “Do the photogrammetric methods conducted for this project efficiently create accurate and accessible 3D models?”, and “How does the use and popularization of photogrammetry for digital recordation change the field of archaeology, and can these changes be utilized in American cultural resource management (CRM)?”. Previous academic work has been done to assess the usefulness and limitations of 3D modeling in archaeology, including archaeological site modeling and individual object modeling. This thesis will build on previously published photogrammetry works, while also documenting specific methods used for the creation of 3D models to produce a photogrammetry “toolkit” for archaeologists to employ on projects. Additionally, this paper will detail the benefits and challenges associated with carrying out photogrammetry for ongoing archaeological investigations
ECOLOGICAL INTENSIFICATION WITHIN FORAGE SYSTEMS BENEFITS SOIL ARTHROPODS AND SOIL BIOTA-MEDIATED ECOSYSTEM SERVICES
No full textModern agricultural practices, such as intensive soil tillage, crop monocultures, andoverfertilization pose sustainability challenges in forage and livestock farming, impacting soil
quality and ecosystem stability. Throughout the world, studies have revealed that agricultural
intensification negatively impacts aboveground arthropods, prompting interest in ecologically
intensified forage systems for agricultural sustainability. Belowground, soil arthropods provide
many essential ecosystem services on the farm, including decomposition, biological control, and
bioturbation; however, these organisms and their services are generally overlooked.
Understanding the response of soil arthropods and soil properties to changes in land
use/agricultural management practices is vital for ecologically and economically balanced
systems. To investigate how land use in forage systems influences soil arthropod communities,
soil properties, and their ecosystem services, I sampled soil biotic and abiotic properties and
measured the rate of feeding activity by soil biota. This study was conducted across a land use
gradient ranging from intensively managed to semi-natural habitats, including established cornsoybean rotations, forage pastures, grass margins, and woodlots. I found significant differences
in several soil properties including soil bulk density and soil moisture between land use types.
Soil arthropod communities were more abundant and diverse in pastures compared to corn plots
and the soil biological quality was significantly lower in corn plots compared to all other land
uses. Soil taxa such as Acari, Collembola, Diplopoda, and Chilopoda were associated with soil
properties investigated here and soil biota feeding activity was highest in ecologically intensified
land use types (characterized by high plant diversity, plant perenniality, and system circularity).
The results of this study suggest that ecological intensification, through the presence of plant
diversity, perenniality, and system circularity, supports soil quality and soil arthropod
communities within forage systems. This research informs decision-making in livestock systems
as they continue to dominate land use throughout the United States
PALS 2025 : Deliverable 1A-5
No full textThis report serves as FTA TOD Planning Grant deliverable 1A-3, a Policy and Process Framework for Equitable Placemaking. It introduces key ways arts-and-culture interventions can help achieve TOD planning goals under this grant, which align with the Purple Line Corridor Coalition’s (PLCC’s) defined fourth priority outlined in the 2017 Community Development Agreement: “Vibrant and sustainable communities enhance health, culture, and a sense of place.” The report outlines how PLCC, leveraging its partnerships with communities, public agencies, and faculty and student design capacity within the University of Maryland, can support future creative placemaking work along the Purple Line corridor in ways that reflect and fortify the strong identities of the neighborhoods along the line.Acknowledgements: School of Architecture and Urban Planning, Creative Placemaking Minor, PALS (Partnership for Action Learning in Sustainability), Purple Line Corridor Coalition, Takoma Langley Crossroads Development Authorityhttps://drive.google.com/drive/u/1/folders/14XAAm174B6Pt3Q1IwTZrMz3Q1NBTw64
THERAPISTS’ RESPONSE TO PROSPECTIVE CLIENTS: THE ROLE OF PERCEIVED GENDERED, RELIGIOUS IDENTITY ON THERAPISTS’ ELECTRONIC RESPONSE TO NEW CLIENTS
No full textMuslim American adults experience significant discrimination (approx. 60%) in personal and institutional settings in part due to disproportionately negative media portrayals that racialize Muslim Americans in gendered ways (Joshi, 2006; Karim & Eid, 2012; Mogahed & Ikramullah, 2022; Shaheen, 2003; Zaal, 2012). Despite this, no comprehensive data exists on the extent to which Muslim Americans can access mental health care (Outadi & Bedi, 2024). This is concerning in light of recent findings that biases based on religion, gender, socioeconomic class, and sexual orientation influence the way mental health clinicians respond to individuals seeking counseling or psychotherapy (Kugelmass, 2016, 2019; Moscovitz et al., 2023; Outadi & Bedi, 2024; Shin et al., 2016, 2021). The current study investigated whether prospective clients are likely to experience differential treatment when seeking mental health services based on their gendered, religious identities. A sample of clinicians (N=981) received help-seeking emails from one of four fictitious clients (Muslim woman, Muslim man, Christian woman, Christian man). Findings provide potential evidence for gendered, religious bias as well as overall gender bias among mental health care providers across the United States. Results demonstrated that across all primary measures (responsiveness, receptiveness, time to respond), one of the two Muslim American confederates received significantly less access to mental healthcare services compared to a Christian American confederate, with the Christian woman emerging as the most likely to be favored and the Muslim man emerging as the least likely to be favored. Furthermore, women confederates, regardless of religion, received more access to services than men. Recommendations for future clinical practice, training, and research are provided