Environmental and Occupational Health Sciences Institute
Rutgers University Community RepositoryNot a member yet
58345 research outputs found
Sort by
Looking backward & looking forward: process photography in the United States, ca. 1970
No full textThis dissertation explores American photography from the 1960s and 1970s through the critical lens of process. By process, I refer specifically to the material and chemical steps involved in the creation of a photographic object. The photographs discussed herein were made in the darkroom just as much, if not more, than in the outside world. I call such objects process photographs, a term of my own creation. For several decades prior to the spread of process photography in the sixties, art photography was virtually synonymous with “straight” photography, an aesthetic doctrine that restricted the materials and techniques a photographer could use in the darkroom on the basis of maintaining medium-specific purity. Yet amid the upheavals of the era, young photographers began to reject such modernist values. Some looked back to the medium’s unruly nineteenth-century past, reviving antiquated processes that predated the industrial standardization of photographic materials. Others experimented with new imaging technologies, tinkering with their mechanical systems and exploring their untapped potential to create art.
They produced a rich and diverse corpus of images, which I analyze through the themes of alchemy, interruption, weaving, and aftereffects. Chapter 1 is organized around the theme of alchemy, a process through which different elements are synthesized to yield a new object. Process photographers worked like alchemists when combination printing, solarizing and chemically staining prints, and employing other transformative darkroom effects. Their photographs tend to invite highly subjective interpretations, which I connect to Carl Jung’s conception of psychological alchemy and the spiritual esotericism of sixties counterculture. Chapter 2 examines photographers who manually interrupted the photographic process. The resulting images often look spoiled, or somehow off, thereby interrupting the window-onto-the-world illusionism of photography and the passive viewing habits it encourages. Interruptions amounted to glitches that ran counter to the supposed rationality and infallibility of Cold War technology. Chapter 3 explores the work of process photographers who incorporated manual techniques, such as sewing and liquid emulsions, into their processes, thereby drawing attention to the hand, rather than the eye, of the photographer. Their work represented a weaving of new methods and subjectivities into the discursive fabric of photography. Chapter 4 considers aftereffect techniques, such as overpainting and hand-tinting, that occur after the initial photographic process has been completed. It closes by exploring the historical aftereffects of process photography through a study of contemporary artists whose works echo the central themes of this project.Ph.D.Includes bibliographical reference
The sound of morals: figures of grammar and music in Dante
No full textThis dissertation explores the relationship between music and grammar in Dante’s linguistic and poetic theory. I argue that Dante departs from a grammatical study of language rooted in Latin and artificial rationality (Dante considered Latin to be an invention of human beings). I instead propose that his study of language is informed by the philosophical study of musical ratios. By adopting a musical model to rethink language, Dante can establish a close connection between language and ethics. In fact, the textual evidence suggests that he applies to the sounds of language the same properties that Greek philosophers had applied to music: a beautiful sounding language will lead to refined morals, and, conversely, a language that features ugly combinations of phonemes will cause its speakers to have equally despicable customs.
Thus, I contend that the public poet envisioned by Dante is also a musicus: he knows the cosmic ratios that bind together the universe and are observable in music. Such knowledge allows poets to reform the natural language of a community and consequently its customs. This enormous ethical authority that Dante ascribes to poets derives from their musical, that is to say, metaphysical, knowledge of language; in fact, the acoustic ratios that they use to mold language are the same cosmic ratios that God established to create harmony in the universe. In other words, I posit that Dante is attempting to find a metaphysical foundation for the ethical reform that he envisions for Italy by applying musical rules to the natural vernacular. This itinerary taking him from esthetics to ethics, and ultimately to metaphysics, is carefully and gradually developed in both his theoretical works, the De Vulgari Eloquentia (On the Vulgar Eloquence) and Convivio (The Banquet), as well as in the Comedy.
In Chapter One, I situate Dante’s use of grammar and music as frameworks of linguistic inquiry in the broader context of late ancient and medieval poetic criticism. I analyze Dante’s De Vulgari Eloquentia and Convivio to provide evidence that his original assimilation of musica speculativa allows him to root poetic authority in metaphysics. In Chapters Two to Four, I show how at key junctures of his reflection on the civic role of poetry and poets in the Comedy, Dante yokes together what I termed “figures of grammar” and “figures of music.” I map this twofold investigation into the grammatical and musical components of language throughout the Comedy to provide new insights on Dante’s linguistic, ethical, and political theory of poetry.Ph.D.Includes bibliographical reference
DNA transfer as a driver of eukaryote genome evolution
No full textAs molecular and sequencing technologies have advanced, many of the biological phenomena we consider to be canonical have become more circumspect, riddled with stipulations, qualifications, exceptions, and often, a broader context. One such process is horizontal genetic transfer (HGT), or the acquisition of foreign DNA by a genome within a generation. This process was traditionally thought to be unique to the prokaryotic domains of life (Bacteria, Archaea) where it is common. However, it is now widely accepted that HGT is a significant mechanism for driving genetic variation, adaptation, and evolution in eukaryotes as well. Conspicuous cases of adaptive (often ancient) HGT identified in genomes of, typically model organisms, have established the study of eukaryotic HGT as a worthy pursuit due to the striking and relatively instantaneous functional innovations conferred by protein-coding genes that have persisted within genomes and outlasted selective pressures. Yet, further investigation must be made into properties underpinning this process, such as frequency, HGT diversity (i.e., coding vs. non-coding), transfer mechanisms, and ecological drivers. Furthermore, these principles must be investigated in non-model systems to elucidate the true scope of this process within the eukaryotic domain. Through the work in this dissertation, I address these important topics. First, I evaluate the current state of the field to consolidate disparate quantifications of HGT within microbial eukaryote genomes and provide ecological context in cases of adaptive HGT. Second, I functionally validate HGTs encoding heavy metal detoxification in extremophilic Cyanidiophyceae red algae and use the findings to explore mechanisms of persistence and integration into new host genomes. Third, I present an alignment-free method for phylogenetic characterization of rare, cryptic, or undescribed taxa from environmental genomic datasets that is effective across a broad range of assembly qualities; this work advances knowledge about non-model organisms that is highly applicable to studies of HGT. Last, I adapt another k-mer-based bioinformatic method for HGT discovery that identifies recent, non-coding, or relic DNA transfers from metagenomic data to gain insights into the frequency and quality of HGTs across a single geothermal springs microbial community.Ph.D.Includes bibliographical reference
Biology, ecology, and management of the Asian longhorned tick, Haemaphysalis longicornis, in New Jersey
No full textThe Asian longhorned tick, Haemaphysalis longicornis is an invasive tick species native to eastern Asia and currently present in parts of eastern and central US. Its parthenogenetic mode of reproduction enables rapid population growth in new areas. In the U.S., it has been implicated in the transmission of pathogens to cattle, but its public health significance is less clear. H. longicornis is not highly anthropophilic, yet it can reach extremely high densities in public parks and is a known vector of human pathogens in its native range. The primary goals of my dissertation are to review the potential impacts of this tick on the health of humans and domestic animals and to develop management strategies focused on the biology of the species while minimizing nontarget impacts. In doing so, I investigate its susceptibility to acaricides and the potential for acaricide resistance and provide a comprehensive analysis of the seasonal dynamics of H. longicornis in New Jersey.
In Chapter 1, I document the first report of multiple human bites by larvae of Haemaphysalis longicornis Neumann in the U.S. Exposure occurred in October in a public park with an extremely high tick population. Although it is not an anthropophilic species, high populations of this tick can increase exposure risk to humans. Despite its ability for rapid population growth and its ability to vector human and animal pathogens, current research on controlling H. longicornis in public spaces is limited. In Chapter 2, I assessed the efficacy of a pyrethroid acaricide (lambda-cyhalothrin) applied in a public park at different intervals according to the seasonal activity of different life stages. Nymphs were most abundant in the spring, followed by adults in mid-summer, and larvae in the fall. Applications of lambda-cyhalothrin timed at the peak of each life stage provided 100% control for 5-7 weeks. Ticks re-established after each application, however, at greatly reduced levels in the mid and late summer applications. To further explore H. longicornis control strategies, in Chapter 3 I evaluated acaricides with different modes of action (pyrethroids, a carbamate, and insect growth regulators [IGR’s]) and delivery methods (spray & granule) to control H. longicornis in an infested park, as well as nontarget impacts of these acaricides on soil and above-ground invertebrate communities. Pyrethroids were the most effective control agents by far and controlled > 95% of the population for the duration of the year. However, pyrethroid applications prompted enduring changes in soil-dwelling arthropod communities, which failed to recover during our survey period. Further research is needed to identify effective and environmentally sustainable alternatives.
Acaricide resistance is a common problem worldwide for numerous tick species that infest livestock. However, baseline acaricide susceptibility, which is critical for resistance determination, has not previously been reported for H. longicornis in the U.S. or elsewhere. In Chapter 4, I examined the susceptibility of a U.S. H. longicornis population to three different acaricide classes using the standard larval packet test. H. longicornis was highly susceptible to propoxur, carbaryl, and coumaphos, and moderately susceptible to permethrin. I concluded that acaricide resistance is not presently a concern for H. longicornis, however, responsible management and resistance monitoring should be employed to ensure the sustainability of tick control products.
As reported in Chapter 2, the phenology of H. longicornis shows considerable overlap in life stages throughout the year likely attributed to the presence of different cohorts. In Chapter 5, I demonstrate in laboratory studies that total lipids provide a good approximation for the physiological condition in H. longicornis, and subsequently measure lipids in field-collected H. longicornis from March to October to give insight into the overlapping generations in its life cycle. Overall this analysis identified two cohorts of nymphs which overlap in the spring, two cohorts of adults which overlap in the spring and summer, and two cohorts of larvae which do not overlap during the year. These results help explain the complex seasonality of the species and indicate that all stages overwinter in New Jersey. Ph.D.Includes bibliographical reference
Literature and the cosmopolitical imagination: reading with others for ecological justice
No full textMy dissertation, “Literature and the Cosmopolitical Imagination: Reading with Others for Ecological Justice,” starts from the premise that the first step in instilling an ecological ethics of care to counter environmental injustice is to recognize all relevant parties, diverse humans and nonhumans alike, as stakeholders in the world that we share. I call this a “cosmopolitical imagination”: one in which both human and nonhuman entities are conceived of as actors and political stakeholders to whom care is due. The dissertation focuses on how literature can spur this imagination by representing—portraying and advocating for—diverse human interests alongside nonhuman ones, catalyzing values and choices about how we relate to others. This project is both interdisciplinary and multilingual: I read fictional narratives from a breadth of cultural and geographic origins—in accordance with the global yet unequal reach of climate change—in order to counter the culturally specific beliefs about animals and the nonhuman environment that drive environmental destruction and block change. In particular, I show how the literary strategies of nonhuman narrators, fictional genres such as speculative fiction and horror, figurative language, and especially the political action of reading with others elicit engaged readers’ care for entangled human and nonhuman interests. Conversations with my students guide my choice of texts, and I draw from their readings in order to foreground the transformative effects on beliefs and behaviors of sharing imaginative literature with others. In this way, just as my project destabilizes cultural and environmental hierarchies in order to imagine different ways humans can relate to the more-than-human world, it also recognizes new, diverse and vital sources of knowledge—both fiction and collective reading—to spur action on environmental injustice. This opens up possibilities for both caring and con/fabulation: building new stories and new worlds, together.Ph.D.Includes bibliographical reference
New GPU-accelerated free energy simulation methods in AMBER
No full textThe drug development process, spanning over 10 years and costing more than $1 billion, involves complex stages from discovery to clinical trials. Structure-based drug design (SBDD), particularly utilizing computational methods, and in particular, molecular simulations, plays a crucial role in streamlining early phases, predicting binding affinity, and aiding lead compound optimization. Despite recent advancements, challenges persist in achieving robust and affordable tools for high-precision free energy predictions in drug discovery applications. In this dissertation, several improved methods have been developed for accurate and robust alchemical free energy simulations, and apply these methods to the prediction of protein-ligand binding affinities of drug targets. Chapter 2 outlines the general background and approaches for absolute/relative free energy calculations. Chapter 3 provides the comparison of the consistency of alchemical transformations of relative and absolute hydration free energies via the λ-dependence of 1-4 van der Waals and electrostatic interactions between the softcore and common core atoms for general cases. In Chapter 4, the new softcore potential form is developed which combined with smoothly transforming nonlinear λ weights for mixing specific potential energy terms, along with flexible λ-scheduling features, to enable robust and stable alchemical transformation pathways. Chapter 5 presents the improvement of the precision of RBFE calculations with an alchemical enhanced sampling approach (ACES) on a large and diverse set of proteins and small moleucle ligands. Chapter 6 illustrates the combination of ACES and the λ-dependent Boresch restraint, and applied on the core-hopping in drug discovery.Ph.D.Includes bibliographical reference
Enhancing efficiency and accuracy in virtual metrology for semiconductors with innovative sparse dimensionality reduction and advanced ensemble learning for multi-source data analysis
No full textIn the semiconductor industry, virtual metrology (VM) serves as a cost-effective and efficient strategy for monitoring processes across different wafers. This approach involves creating a predictive model that leverages real-time data from equipment sensors alongside measured wafer quality characteristics. Before crafting this prediction model, it's crucial to carefully select relevant input variables to ensure the subsequent analysis remains efficient, despite the high dimensionality of sensor data. However, the extensive use of multiple sensors in wafer production can lead to increased costs. Principal component analysis (PCA) is crucial for extracting meaningful information from these large datasets, enabling the identification of key variables that reveal underlying patterns. As the number of variables increases, the sparse PCA (SPCA) becomes essential for dimensionality reduction and improving interpretability. However, existing SPCA methods often fail to identify a small but highly influential subset of input variables, reducing the effectiveness of the analysis and potential cost savings.
Firstly, this dissertation introduces new technique that address various aspects of sparse principal components (PCs), exploring a spectrum of SPCA variants. We unveil a variant named true sparse PCA (TSPCA), which aims to use a small subset of input variables in the initial PCs, thereby reducing the need for numerous sensors. Our experiments demonstrate that TSPCA not only lessens sensor dependency but also provides a comparable level of variance explanation to existing SPCA methods.
Moreover, to counter the inefficiencies of traditional enumeration algorithms with large-scale data, we propose the dynamic TSPCA (D-TSPCA), a high-efficiency algorithm designed for TSPCA to solve high-dimensional datasets. D-TSPCA innovatively secures true sparse loading vectors through a floating selection process, overcoming the constraints of previous algorithms by allowing for the dynamic selection and deselection of variables. This method effectively minimizes essential sensor numbers in the first several numbers of PCs while maintaining a similar level of explained variance. Our results demonstrate D-TSPCA's capability in handling complex, high-dimensional problems.
Furthermore, we present the least angle SPCA (LA-SPCA), designed to tackle computational hurdles inherent in SPCA. Utilizing the least angle method, LA-SPCA identifies sparse PCs with minimal deviation from their ordinary PCA counterparts, significantly streamlining computational demands and boosting efficiency in processing ultra-high dimensional datasets. Our comprehensive experiments affirm the superior performance and efficiency of this novel SPCA technique.
Finally, we have developed a data-fusion model to recognize the variability in sensor data characteristics and the potential degradation in model performance from aggregating diverse data sources in VM process. This model integrates multiple data sources into a cohesive multi-source framework, enhancing the AdaBoost regression algorithm for VM applications. By merging residuals from individual and collective data sources to adjust thresholds adaptively, we ensure the accuracy of predictions for each weak learner. Real-world processing data from semiconductor manufacturers have demonstrated that our approach surpasses single learning algorithms in robustness and performance, marking a significant advancement in the field of VM.Ph.D.Includes bibliographical reference
Automated machine learning for intelligent systems
No full textThe fast progress in Machine Learning techniques has led to a growing impact of Artificial Intelligence (AI) on various aspects of people's lives. AI model learning comprises three vital components: data inputs, model design, and loss functions. Each of these components makes a significant contribution to the AI system's performance. Traditionally, these components needed skilled domain experts to meticulously design them, creating a challenging barrier to entry into AI. Besides, manual approaches are relatively inefficient and often fail to achieve optimal results. Recently, automated machine learning (AutoML), such as neural architecture search (NAS), has tried to tackle the challenge by automating the design and parameters of deep models. However, traditional AutoML research mainly focuses on automated model design instead of the inputs and loss functions, and AutoML's research on recent large language models (LLMs) is still in its infancy due to the enormous computations required. Consequently, this thesis aims at answering the following questions:
1. How can we design AutoML techniques for the other two essential parts of AI systems, i.e., inputs and loss functions?
2. What is the role of AutoML under the background of the rapid development of large-scale models such LLMs? How can we leverage the insights of AutoML to improve gls{LLM}s?
Given these questions, this thesis introduces three research works:
1. We propose AutoLossGen [1], which can generate the most effective loss function using basic mathematical operations for various recommendation tasks and models. The generated loss gives better recommendation performance than commonly used baseline losses and is transferable to another model or dataset on the same recommendation task.
2. As for the data input aspect, we create the Personalized Automatic Prompt for Recommendation Language Model (PAP-REC) framework [2]. This framework aims to teach language models the optimal automatic input instructions, i.e., prompts, for recommendation tasks. We develop surrogate metrics and leverage an alternative updating schedule to address the inflated search space derived from personalized prompts. Our PAP-REC framework manages to generate personalized prompts, and the automatically generated prompts outperform manually constructed prompts.
3. For LLM-based agents, our Formal-LLM framework utilizes formal language through automaton to automate the workflow of LLM-based agents when solving complex tasks under constraints [3]. Our framework enables the agents to automatically adhere to the planning policy with automata guidance, resulting in more controllable LLM-based agents.
In these three studies, through the development of automated machine learning techniques at various stages of an AI pipeline, we further advance both small-scale and large-scale intelligent systems, making them more precise, effective and user-friendly.Ph.D.Includes bibliographical reference
Prioritizing federal investments for coastal adaptation
No full textClimate change is having widespread impacts around the world. The coastal zone is a dynamic place to study how humans collectively respond and adapt. Drawn by economic opportunities and coastal amenities, billions of people have settled in low lying areas exposed to sea level rise and increasingly intense storms. Coastal hazards pose an existential threat to many communities and will likely force mass migration and displacement in the coming decades. At the same time, many communities are resisting relocation despite repetitive flood losses. Given the diffuse nature of coastal hazards, large upfront costs for adapting physical infrastructure, and competing interests across spatial and temporal scales, conflicts emerge about the proper course of action in diverse local contexts.
This dissertation explores the political economy of coastal climate adaptation, or the struggle for power and resources between competing interest groups. While bringing together various literatures, the work is primarily situated in the field of planning and public policy, with a focus on fiscal federalism: I explore the multi-level governance arrangements that shape and distribute coastal risk through various policy choices. Via case studies in the United States, I examine the programmatic and policy mechanisms through which federal agencies distribute resources for coastal flood risk reduction. I study how values, beliefs, and worldviews inform preferences for coastal adaptation strategies, and evaluate tradeoffs from diverse perspectives.
Findings identify large, uncoordinated intergovernmental transfers as a key issue with current fiscal arrangements: in certain post-disaster contexts, these transfers enable policy capture by elite local and private interests. Federal investments in supporting and protective infrastructure create the market conditions that encourage reinvestment in areas experiencing repetitive flood losses. By artificially softening local government and household budget constraints, these fiscal transfers create market distortions, incentivize moral hazard, and externalize costs to non-coastal communities and future generations. To address these issues, policy recommendations include to:
1)develop a national coastal land use strategy,
2)encourage mobility away from the coast at scale,
3)incrementally adjust methods for characterizing the tradeoffs of adaptation strategies, and
4)move toward transformative adaptation by redressing the drivers of climate vulnerability.Ph.D.Includes bibliographical reference
Promoting the social and emotional well-being of middle school students in visual arts classrooms
No full textEducators are increasingly recognizing the importance of social-emotional learning (SEL) as a response to the social and emotional challenges faced by students in the aftermath of the COVID-19 pandemic. In alignment with this, the state of New Jersey has mandated the explicit integration of SEL into K-12 art curricula, effective from Fall 2022. While many arts educators have acknowledged the implicit occurrence of SEL in art classes due to the nature of artistic activities and lessons – such as fostering creativity, self-expression, content creation, and interpersonal connections – there is a lack of empirical research on the intentional and explicit integration of arts and SEL curricula to inform best practices in the art classroom. Consequently, the current study aims to address this gap by piloting the intentional embedding of SEL into the visual arts curricula of two middle schools within a suburban school district in New Jersey, during the first trimester of the 2022-2023 school year.
Using a sequential explanatory mixed-methods design within a program evaluation framework, the current study incorporated quantitative measures of arts performance and indices of students’ mental health, along with qualitative interviews with teachers and classroom observers. Quantitative analyses (i.e., analyses of covariance and regression analyses) examined differences among students in SEL-embedded art classrooms, while thematic analyses of interview transcripts complemented and expanded upon the quantitative findings. Although significant changes were not found in any of the indices of students’ mental health from pretest to posttest in either of the schools, students’ self-efficacy, SEL skills, and satisfaction in the course were significantly associated with improved artistic performance – as measured by final art grades. Qualitative themes emerging from semi-structured interviews provided insights into the implicit and explicit integration of SEL within the visual arts curriculum, as well as barriers to SEL implementation and areas for potential growth. These findings can help inform future development of curricula and programs integrating SEL within the visual arts.Ph.D.Includes bibliographical reference