50107 research outputs found
Sort by
ASSESSING THE FEEDING PREFERENCES OF THREE EARLY-SEASON INSECT HERBIVORES ACROSS SIX APPLE CULTIVARS
No full textThe overreliance on chemical controls in agricultural production has contributed to widespread insecticide resistance and the development of long-term health complications in humans, emphasizing the need for alternative pest management strategies. Trap cropping and habitat manipulation can be used as integrated pest management (IPM) techniques for insect monitoring and to attract pests to preferred host crops. This study investigated three early-season insect herbivores— including the rosy apple aphid (RAA), Dysaphis plantaginea (Hemiptera: Aphididae), green apple aphid (GAA), Aphis pomi (Hemiptera: Aphididae), and pear thrips Taeniothrips inconsequens (Thysanoptera)— which can cause significant damage to apple tree foliage and crop yields if left uncontrolled. Research was conducted across two Massachusetts orchards during the 2023 and 2024 growing seasons and focused on multi-cultivar grafted apple trees that contained six cultivars (Liberty, Red Astrachan, Ginger Gold, Dabinett, Wickson, and Yellow Transparent) and a Stock branch. Fruit clusters were randomly sampled per cultivar/tree and assessed for three biological parameters: insect incidence, abundance, and parasitism (2023 only, for RAA). The diversity of natural enemies was also recorded from 2023-2024. Results from 2023 revealed that Red Astrachan, Ginger Gold, and the Stock (the original cultivar present on the apple trees) were more attractive to RAA, while Wickson showed consistently higher infestations of GAA and pear thrips in 2024. These findings support the use of multi-cultivar grafted apple trees as part of an ECOstaking approach to insect pest monitoring on apple trees, with the potential to reduce chemical inputs by concentrating pest pressure in targeted areas.Master of Science (M.S.)2026-05-1
Snow Production in Todays Changing Climate
No full text“Snow Production in Todays Changing Climate” is a thesis to explore artificial snow production options. Methods for forecasting weather, including: moving average, weighted moving average, exponential smoothing, regression, are evaluated to explore weathers relevance to operations of snow production, maintenance of snow, reuse of precipitation and snowmelt, solar energy for snow production. Snow production choices are explored all through the lens of minimizing costs with consideration to environmental friendliness. Concepts of cost benefit analysis and breakeven point is used in the method for snow gun choice between three options, Portable Air Water Gun, Standard Low Energy Air Water Gun, and Portable Fan Gun, as well as for the choice of snow farming practices and reservoir building. Linear and exponential regressions are used for optimization in terms of cost and savings functions. The amount of fuel, electricity, or water, saved from usage of resources discussed in this study is an indication to the amount by which a resort could impact various environmental foot prints, whilst being conductive to cost minimization of operations.Master of Science (M.S.
Women in Medicine Digital Choice Board
No full textThis choice board invites students to write a news story about the first American Indian woman to receive a medical degree; create an Instagram-style video showcasing the Blackwell Sisters’ achievements in the field of medicine; Create a graphic novel or picture book about an African American medical pioneer during the Civil War, and much more
Observer-Aware Planning Under Uncertainty
No full textAs more autonomous agents share space with people, it is crucial that these agents are cognizant of how their behaviors are interpreted. Previous work has investigated various methods to communicate or conceal agents' goals, intentions, and capabilities through their behaviors. Many of these methods are based on the same principle: the agent reasons about how its actions will be interpreted by observers and selects actions that lead to desired beliefs in the observer.
Despite the commonality of this principle, these methods are often developed in isolation, focusing on specific aspects of the problem. This fragmentation makes it difficult to compare and combine these methods to create more sophisticated behaviors.
This thesis presents a unifying model for generating behaviors that achieve desired goals while considering how these behaviors are interpreted. The proposed model, referred to as the Observer-Aware Markov Decision Process (OAMDP), operates under the assumption that observers interpret the agent's actions to form beliefs about the agent's potential desires, goals, and intentions. Rewards in OAMDPs depend on the observer's beliefs in addition to the agent's state and action. Using belief-dependent rewards, we demonstrate how OAMDPs can generate behaviors that lead to desirable observer beliefs. Additionally, we present an online user study comparing the observer-aware behaviors generated by OAMDPs to those generated by a baseline model. The results show that observers interpret the OAMDP-generated behaviors as intended more frequently.
OAMDPs provide an expressive framework capable of producing various kinds of observer-aware behaviors. However, reasoning about the observer's beliefs introduces a dependence on histories. We provide a proof that this dependence makes solving OAMDPs intractable in the worst case. To address this issue, we investigate several approximation algorithms and present error bounds for a special case where rewards and belief updates are Lipschitz-continuous. We then empirically evaluate these algorithms' performance on a variety of domains.
Although belief-dependent rewards are useful for generating observer-aware behaviors, they can conflict with task efficiency. To achieve a principled balance between task efficiency and observer-awareness, we introduce Constrained OAMDPs (COAMDPs). The introduction of constraints allows us to specify the desired observer beliefs while enforcing constraints on task efficiency. Since the constraints make solution methods for unconstrained models inapplicable to COAMDPs, we propose new algorithms and demonstrate their effectiveness through empirical results.Doctor of Philosophy (Ph.D.)2026-05-1
Speaking Pacifically: On Being Native in the Digital Age
No full textThis dissertation draws on ethnographic research conducted with Kanaka Maoli (Native Hawaiian) youths between the ages of 18 and 35 that have created and maintain an active presence on Instagram. The research is an examination of the varied possibilities of identity expression and digital community building for Native Hawaiians in diaspora and how connections are maintained and built to structure a digital community in which Indigeneity is a present tense reality of life that combats stereotypical representations rather than a historical concept. I consider specific ways in which social media is utilized as a powerful tool for social expression as well as safe-guarding community in a “post-COVID” social landscape where calls by on-island community members highlight on-the-ground narratives of trauma and social issues that kānaka are facing. I demonstrate multiple ways in which social media, for as many issues as it has, empowers everyday Native folks to tell their stories in unfiltered and unmediated ways.UMass Amherst Graduate School Dissertation Completion FellowshipDoctor of Philosophy (Ph.D.)2026-05-1
Responsive Materials for Regulating Activities of Enzymes and Synthetic Catalysts
No full textEnzymes and synthetic catalysts offer many advantages in synthetic routes for greener processes. For example, enzymes provide high selectivity and specificity, organocatalysts provide robustness in air and moisture, and organometallics provide a wide substrate scope and reactivity based on metal and ligand pairing. Each have their advantages in producing chiral drugs, natural products, and enantiomerically enriched bioactive molecules. Despite their potential, they are inherently limited to their native reaction media and often lack a degree of tunability and selectivity. Expanding their catalytic transformations in non-native environments, such as enzymes in organic solvents, and organometallics in aqueous media, while also providing precise control and regulation in various conditions, will greatly expand the substrate scope and synthetic transformations available in terms of novel reaction pathways and selectivity previously inaccessible. My research is focused on understanding and developing design rules for polymeric scaffolds, supramolecular assemblies, and small molecules that can be programmed to activate or deactivate catalysts in different environments: a) we have engineered random co-polymers that self-assembled into reverse micelles to provide an aqueous core for enhanced enzyme stability in bulk organic solvent. These polymers incorporated hydrophilic PEG and alkyl units to maintain the hydrophilic and lipophilic balance for assembly formation, a ‘charged’ unit to sequester enzymes into the assembly via electrostatic interactions, and a light-responsive unit to introduce crosslinks that control substrate permeability into the assembly. We further explored how variations in the density of the ‘charged’ unit can modulate enzyme activity to select complementary charged substrates, demonstrating unnatural enzyme selectivity. b) We constructed protein-responsive small molecule chimeras and amphiphilic dendrimers to provide synthetic catalyst stability in aqueous media. Each molecule linked a catalyst to a protein ligand either by a hydrophobic chain, which aided in noncovalent encapsulation inside amphiphilic assemblies or covalent attachment to an amphiphilic dendron unit, which would self-assemble to provide a hydrophobic core suitable for catalyst stability. Both systems were studied for catalyst stability and dormancy inside the assemblies and their responsiveness towards non-enzymatic proteins to turn catalysis ‘on’. c) We designed small molecule pro-inhibitors to specifically target and regulate protease activity. These molecules are substrate analogs that remained reversible and inactive until their specific protease acted upon them. The extent of protein inhibition and adduct formation was studied using fluorogenic assays, MALDI-MS, and LC-MS.Army Research Office (W911NF-15-1-0568)
NSF NRT-SMLS Traineeship and Fellowship (DGE-1545399)Doctor of Philosophy (Ph.D.)2026-05-1
Epigenetic Aging of Sperm as a target for Chemical Exposures
No full textEvidence showed that sperm epigenome is affected in similar pattern by various factors such as diets, physical exercise, psychological stress, age, exposure to environmental toxicants. These alterations can impact offspring health leading to impaired metabolic, neurodevelopmental, and psychiatric health. Among all studied factors, aging has the strongest effect on sperm DNA methylation and its effects can be modulated by environmental factors. As an example, studies have shown that chemical exposures can alter the rate of aging of the sperm epigenome. However, the mechanism through which aging, environmental factors, and other stressors affect sperm DNA methylation in similar patterns remains unknown. Therefore, this work tries to understand the possible mechanisms behind this association. Research demonstrates that significant epigenetic reprogramming during spermatogenesis occurs in the biochemical environment regulated by the permeability of the blood-testis barrier (BTB). This permeability is controlled by the balance of the mammalian target of rapamycin (mTOR) complexes (mTORC1 and 2), with mTORC1 promoting disassembly of the BTB and mTORC2 promoting its integrity. Thus, I hypothesized first that the balance between mTOR complexes in the Sertoli cell is responsible for the age dependent changes in sperm epigenome due to diverse stressors. Secondly, the conserved patterns observed in sperm DNA methylation changes by aging and other stressors suggest that these changes are due to the sensitive nature of specific DNA regions.
To test my first hypothesis, mouse model with permanently tight and permanently loose BTB were generated, by carrying out a Sertoli cell-specific inactivation of mTORC1 or mTORC2 via knockout of corresponding components of mTOR complexes, Raptor and Rictor, using Cre-Lox approach. Our findings demonstrate that loosening of the BTB through mTORC2 suppression accelerated sperm epigenetic aging in younger mice while tightening of the BTB through mTORC1 suppression in older mice rejuvenated sperm DNA methylation profile.
The conserved pattern of changes in sperm DNA methylation also suggest that likely there are some DNA methylation regions that are naturally responsive to reprogramming, which make me hypothesize that the common patterns of sperm DNA methylation changes in response to aging and stressors are due to the nature of variable methylation regions (VMRs) sensitive to methylation reprograming, and therefore may support the model of stochastic epigenetic variation. The stochastic epigenetic variation hypothesis predicts that sperm epigenome contains naturally variable methylation regions (VMRs) associated with developmental genes and that sperm methylation changes in response to stressors occur due to the increased methylation variation of these VMRs. Thus, I test predictions from the stochastic epigenetic variation hypothesis using mouse and rat sperm DNA methylation and other -omics data. Results demonstrate that VMRs overlap significantly with stress dependent DMRs and are enriched with similar biological categories, such as embryonic development, neurodevelopment, and metabolism. Increasing methylation variation in VMRs is responsible for the formation of stress-dependent DMRs due to asymmetrical increase in variation of hypo- and hyper-methylated regions. Findings for the second hypothesis, provide for the first time evolutionary and molecular mechanistic explanation of the conserved patterns of DNA methylation change in response to various stressors.
Lastly, based on the findings from the first hypothesis, about the role of the mTOR/BTB cascade in sperm epigenetic aging regulation, I hypothesize that environmental and pharmaceutical interventions targeting mTOR activity and/or BTB permeability will modify rates of epigenetic aging of sperm. Using a sperm epigenetic clock, environmental factors - 34.5°C heat stress (HS) and cadmium exposure resulted in 31.7- and 42.3-days older sperm respectively than same age in controls, while pharmaceutic mTORC1 inhibitors (rapalogs) rejuvenated sperm epigenetic age when compared with the controls. For example, 0.5 and 1 mg/kg body weight of rapamycin administered for the duration of two spermatogenesis cycles resulted in a significant reduction of sperm epigenetic aging by 30 and 41 days respectively.
Findings from this work for the first time identifies the mechanisms responsible for the conserved patterns of sperm DNA methylation changes induced by diverse stressors. This research identifies these mechanisms as a novel adverse outcome pathway, linking xenobiotic exposures and sperm epigenetic aging. Additionally, it opens opportunities for the development of new approaches for sperm epigenome rejuvenation, and develops a mouse sperm epigenetic clock, a tool to be used in future research to evaluate a variety of interventions on sperm epigenetic age. Overall, we identified mechanisms that may be a significant route of transfer of aberrant epigenetic information to the next generation.Doctor of Philosophy (Ph.D.)2030-05-1
Machine Checked Verification of Validation Tests for Seldonian Algorithms
No full textSeldonian algorithms are a class of algorithms that either return a (usually machine learned) model that meets specified safety conditions with specified probability, or return “no solution found”. One fairly natural way to make a Seldonian algorithm is to take an existing algorithm and subject its output models to some safety validation test(s). This validation testing serves as a filter that lets through only models that satisfy the tests, returning “no solution found” (NSF) otherwise.
In this work we produce verified code for a Seldonian algorithm validation test. This involves (1) proving the underlying mathematics for a Hoeffding’s-inequality-based concentration inequality to be used as the basis for a test, (2) writing an algorithm for the validation test in the HOL4 theorem prover, (3) verifying that algorithm’s correctness and tolerance given representation errors of arbitrary precision rationals under operations like square root, (4) embodying that algorithm in code in CakeML’s dialect of SML, (5) and compiling that code with CakeML’s verified compiler.FMitF: Track I: Verified Safe and Fair Machine LearningDoctor of Philosophy (Ph.D.
Financial Institutions and Financial Elites: Mechanisms of Power Accumulation
No full textThis dissertation investigates the mechanisms through which financial elites and financial institutions accumulate power in contemporary capitalism. Drawing from the political economy of finance, it explores how financial actors exert and benefit from influence across regulatory, political, and market domains—consolidating infrastructural, instrumental, and structural forms of power. Through three essays, the dissertation examines how financial institutions are embedded within governance frameworks that advance financial interests, how financial elites invest in connections to secure favorable outcomes, and how tighter financial regulation in overfinancialized economies can enhance economic growth—challenging academic narratives that have long sustained finance’s expansion.
The first essay analyzes the infrastructural power of finance through the case of the European Banking Union’s Single Resolution Mechanism (SRM). Although formally designed to promote financial stability, the SRM steers interventions toward private market solutions, leading to greater banking consolidation and marketization, particularly in bank-based systems. By prioritizing subordinated debt, limiting public intervention, and embedding high levels of discretion in resolution practices, the SRM reshapes the European banking landscape in favor of large, market-oriented institutions—entrenching financial interests in EU governance.
The second essay examines the instrumental power of financial elites by comparing traditional finance actors with the crypto elite. Using a novel dataset of board members from the 50 largest US-registered crypto firms, the top 50 US financial firms, Fortune 1000 companies, and policy-planning organizations, it applies cluster and network analysis to compare socio-educational traits, corporate affiliations, and political connections. Results show that the crypto elite closely mirrors traditional finance in background and power investments—such as political donations and revolving doors—while occupying a more peripheral position in the US business network. The White House emerges as a key site of past employment for both elites.
The third essay empirically assesses the relationship between macroprudential policies and growth. Using a Local Projections methodology on a panel of Advanced and Emerging Market Economies (2000–2017), it finds that tighter policies are associated with higher GDP growth in overfinancialized economies, with highly liberalized and intermediated financial sectors. No significant effect is found in Emerging Markets. Supporting the “too much finance” literature, the essay contributes to efforts to curb finance’s size and structural importance.Doctor of Philosophy (Ph.D.