50107 research outputs found
Sort by
Accelerating Sustainability of the Electricity Grid using Distributed Energy Resources
No full textIn recent years, the impacts of climate change have become more visible, raising concern and active movement in sustainability efforts. For instance, energy transition, which focuses on shifting from traditional fossil fuels towards renewable energy sources, is a critical strategy for mitigating the effects of climate change. Electrical grid is an important part of the energy transition since it still heavily relies on dirty sources such as coal, oil, and natural gas in many locations. Moreover, other sectors such as transportation, industry, and agriculture are transiting to all electric economy to reduce their emissions which leads to higher demand and, of course, emissions from the grid. The advances of the Internet of Things (IoT) and the proliferation of high-capacity networked energy devices at household-level, such as electric vehicles (EV), batteries, and heating, ventilation, and air conditioning (HVAC) have introduced opportunities for transforming electric demand at house-level and for coordinated control of those residential loads at a large scale. This provides a new and powerful form of demand response in terms of environmental and consumer perspectives to accelerate the sustainability of the grid.
This thesis puts forth a central focus on sustainability in electricity grids with the presence of distributed energy resources. At the same time, the study also takes a human-centric design approach that considers the environmental, economic, convenience, and privacy aspects of electric consumers in the design of the systems and algorithms. To address those challenges, first, I propose a grid peak shaving framework that consists of peak prediction and a control algorithm which utilizes a distributed and heterogeneous pool of energy resources to perform flexible grid peak shaving. The algorithm can take home owner’s preference into consideration. Second, I examine electricity grid peak patterns, and then present peak prediction algorithms that can predict peak time of the day, peak day of the month, and peak day of the year respectively. I also provide reference datasets for peak forecasting in energy systems. Third, to prevent privacy leakage from the electric consumption data, I introduce an algorithm that shifts electricity demand with household batteries to prevent occupancy leakage while preserving other useful information. Finally, I analyze the potential conflict between electricity prices and carbon emissions and the resulting trade-offs in carbon-aware and cost-aware load scheduling. I also present a control algorithm that balances between reducing carbon and cost while still respecting user and grid constraints.Doctor of Philosophy (Ph.D.
LEARNING FROM THE PAST TO INFORM FUTURE ACTION: LEARNING CURVE INNOVATIONS
No full textWe present the work in three essays in the manuscript. Each essay will include methodology and results. Our work will focus on renewable energy technologies and water recycling technologies. Estimating the future cost of technologies that play a significant role in the energy transition can help design robust and cost-effective policies to promote a carbon-neutral, sustainable economic system. These forecasts are important inputs to a number of analyses, including energy-economic models.
For the first essay, we develop a hypothesis around the impact of a related technology on the development of an experience curve. We explore the implications of this hypothesis in the case of wind energy, which has been historically developed onshore and is currently experiencing rapid growth in deployment offshore. We look at the impact of modeling offshore wind as (1) a fully new technology, (2) a direct offshoot of onshore wind, and (3) a hybrid. Focusing on the levelized cost of electricity of offshore wind, we find that assumptions about its relatedness to onshore wind are equally important as assumptions about future growth scenarios. This research highlights a previously neglected factor in experience curve analysis, which may be especially important for technologies, such as offshore wind energy, that are expected to contribute significantly to climate change mitigation.
For the second essay, we investigate in more detail data underlying learning curves. We disaggregate the error between projected technology costs and realized costs, to determine how much of the error comes from projecting cumulative experience and how much from projecting the amount of learning in response to cumulative experience. We apply this method to energy technologies, using actual observed data, projections based on authoritative reports, and expert forecasts. The disaggregation analyses can tell us how much of a surprise in cost is due to higher or lower than expected increases in cumulative experience in the form of installed capacity; or higher or lower than expected learning per unit capacity.
Lastly, for the third essay, we investigate the choice of metrics used in learning curves. We apply this analysis to learning curves of water recycling technologies. This pilot study and analysis, which focuses on the impact of the metric choices in learning curves, will build knowledge for a larger study in the future. We systematically examine a range of studies to gain insights into the metrics utilized for experience and performance in assessing the learning rates of energy technologies. Existing literature broadly suggests that energy technologies that are more investment-reliant rather than process and labor reliant are better represented by cumulative capacity. We develop and focus on a case study for California; assemble a dataset of costs and capacities; and construct an experience curve of water recycling in wastewater treatment plants. We illustrate how metric choice impacts the shape and forecasts of the learning curves.Doctor of Philosophy (Ph.D.
Uncertainty-Aware Computer Vision in Resource-Constrained Environments
No full textRecent breakthroughs in deep learning techniques have led to staggering performance improvements in many domains. This has made autonomous systems a critical component for many real-world use cases, including in Internet of Things (IoT) environments. This domain is especially challenging as it imposes considerable environmental, networking, and hardware constraints on the models. Further compounding this challenge are the high stakes decisions that are necessary in many real-world deployments. This thesis explores how to leverage uncertainty awareness to create more robust vision models for use in constrained environments. By estimating and communicating the uncertainty that a model has, we can generate more reliable and trust-worthy predictions.
We first consider the problem of zero-shot image classification, where no labeled data is available for some classes. By utilizing a textual class hierarchy, we expose an accuracy-specificity trade-off that lets systems make more accurate, albeit less specific, predictions under uncertainty due to resource constraints. We then address the distributed execution of image classifiers. We split a neural network between an edge device and the cloud by performing a partial execution on the edge and sending latent features to the cloud for completion. We find that this approach demonstrates superior bandwidth utilization over conventional methods. Merging these strategies together, we craft a distributed, hierarchical object detector validated via a prototype on ultra low-power edge hardware.
We next evaluate the edge runtime of recent transformer-based object detectors. We additionally show how their unique characteristics simplify reasoning about bounding box uncertainty compared to earlier methods. We approximate the Bayesian parameter uncertainty using a simple deep ensemble. Due to the high cost of this approach, we present a more efficient uncertainty quantification method by ensembling only a subset of the detector's parameters.
However, reasoning about uncertainty over bounding boxes remains challenging and makes multi-camera fusion less straightforward. For these reasons, we consider geospatial tracking, where 3D points in a shared world space are predicted rather than boxes in the image plane. With the support of multi-camera datasets with geospatial ground truth, we train a deep probabilistic model of an object's position. The predictions are then fused using multi-observation Kalman trackers. We demonstrate how modeling the geometric transformation between the image plane and the world coordinate frame allows us to train geospatial detectors for tracking using much less data than end-to-end deep learning approaches. Furthermore, we are able to output intuitive geospatial uncertainty estimates, generalize to unseen viewpoints, and provide straightforward support for multi-object tracking.Research reported in this thesis was sponsored in part by the CCDC Army Research Laboratory under Cooperative Agreement W911NF-17-2-0196 (ARL IoBT CRA). The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Laboratory or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein.Doctor of Philosophy (Ph.D.
Generics and Quantified Generalizations: Asymmetry Effects and Strategic Communicators
No full textGeneric statements (‘Tigers have stripes’) are pervasive and developmentally early-emerging modes of generalization with a distinctive linguistic profile. Previous experimental work suggests that generics display a unique asymmetry between the prevalence levels required to accept them and the prevalence levels typically implied by their use. This asymmetry effect is thought to have serious social consequences: if speakers use socially problematic generics based on prevalence levels that are systematically lower than what is typically inferred by their recipients, then using generics will likely exacerbate social stereotypes and biases. This paper presents evidence against the popular hypothesis that this asymmetry effect is unique to generics. Correcting for various shortcomings of previous studies, we found a generalized asymmetry effect across generics and various kinds of explicitly quantified statements (‘most’, ‘some’, ‘typically’, ‘usually’). In addition, to better understand the conditions under which generalized asymmetry effects may exacerbate biases, we examine whether speakers choose generalizing sentences based simply on their acceptance conditions, or are systematically sensitive to the implications likely drawn by their typical recipients. In support of the latter view, we found that, in neutral or cooperative scenarios, speakers reliably choose generalizing sentences whose implied prevalence levels closely match the actual ones. In non-cooperative scenarios, many speakers exploit asymmetry effects to further their own goals by choosing generalizing sentences that are strictly true but likely to mislead their recipients. These results refine our understanding of the source of asymmetry effects and the conditions under which they may introduce biased beliefs into social networks
Module 7
No full textThis module contains materials on data creation and editing concepts and practice, and a brief exercise reviewing module 4 concepts
A War Story: World War II, Memory, and Experience
No full textThis paper explores World War II and American collective memory in the video game Call of Duty: World at War, (Activision 2008) and how it influences public understanding of the conflict. Drawing on oral histories as well as historical scholarship, the paper analyzes game missions in an effort to discover World at War’s historical fidelity. The findings reveal that many of the game’s missions remain faithful real world locations and dates. While the game encourages historical empathy and moral reflection, its portrayal omits key racial and ethical complexities, particularly in the Pacific theater. This omission reinforces narratives of American exceptionalism when compared to the game’s depiction of the Eastern Front. Nevertheless, World at War offers a unique digital space where players can engage with the trauma and memory of World War II, highlighting the potential and limits of gaming as a medium for historical representation
Turning The Signal Back To Heaven
No full textThis is a collection of poems.Master of Fine Arts (M.F.A.)2030-09-0
Calibrating Trust in Visualization through the Manipulation of Visual Complexity
No full textData visualizations are widely used to understand, communicate, and inform decisions in various fields including healthcare, environmental science, and education. The design choices made when creating visualizations can significantly shape people’s interpretation of data and, hence, their decisions. This raises critical issues surrounding trust in data visualizations, particularly ensuring viewers can accurately gauge the reliability of the information presented. Critical information can be discounted or dismissed without mutual trust between the viewer and the visualization. Therefore, establishing trust is a critical first step in visual data communication. However, existing visualization research has yet to reconcile a model of trust in visual data communication. My dissertation will explore the relationship between visualization design and trust to build a model that quantifies the impact of perceptual factors on trust in visual data communication.
To this end, I comprehensively surveyed literature across social science and computer science and established a multidimensional framework for operationalizing trust in visualization. This framework proposes that trust results from cognitive factors, based on logical reasoning, or affective, driven by emotions and beliefs. The framework further divides trust in visual data communication into two aspects: trust in the quality of the underlying data and trust in the design of the visualization, including its clarity and potential to mislead. Building on this framework, I conducted a series of experiments that identified visual complexity as a key factor influencing both trust in the underlying data and trust in the visualization design. For my dissertation, I will operationalize visual complexity in the context of visualization design and systematically examine its effect on trusting behaviors.
Psychologists have identified processing fluency, the speed and accuracy with which we perceive and interpret a stimulus, as a driving factor of visual complexity. Many perceptual elements can impact the processing fluency of a visualization, including the choice of color, the size of visual marks, the scale of the visualization, and the amount of information displayed. I propose to 1) establish the design space of factors that impact the processing fluency of visualizations, 2) generate a database of visualizations with varying levels of processing fluency through formative studies, and 3) test the effect of fluency on trusting behaviors. These efforts will contribute to a comprehensive set of guidelines for creating perceptually fluent visualizations, with a consideration of how individual design choices can change perceived visualization complexity and impact trust in visual data communication.Doctor of Philosophy (Ph.D.
University of Massachusetts Amherst Open Access Policy Focus Group Protocol
No full textThis is the script used for 4 focus groups about the UMass Amherst open access policy conducted in January and February 2025 with 17 total participants
Assessing Hunter Values, Expectations, and Satisfaction Regarding Controlled White-tailed Deer Hunts in Suburban Eastern Massachusetts
No full textThe Massachusetts Division of Fisheries and Wildlife (MassWildlife), the state agency responsible for managing white-tailed deer (Odocoileus virginianus), is tasked with maintaining deer populations at both biologically and culturally beneficial levels and supports deer management in the form of regulated hunting as a fiscally responsible, ecologically sound, and socially beneficial strategy for meeting that objective. White-tailed deer are ubiquitous throughout the Commonwealth of Massachusetts, particularly in eastern portions of the state where
suburban neighborhoods, scattered woodlots, and ample travel corridors create a landscape mosaic comprised of high-quality deer habitat and protection from hunting. Given that native predators of deer, gray wolves (Canis lupus) and mountain lions (Puma concolor), have been locally extirpated, deer mortality rates due to predation are lower than historic levels. Regulated hunting by licensed individuals serves as one of the only remaining mortality factors for adult deer. Suburban environments, however, often serve as refugia from hunters as firearm discharge setbacks and numerous local bylaws highly restrict hunter access to suburban deer habitat. Lack of predation by both extirpated carnivores and hunters coupled with the abundance of high-quality habitat have resulted in increased survival and reproductive rates of suburban deer. Subsequently, deer populations in suburban eastern Massachusetts have continued to grow and become denser over the long-term. High deer densities in these heavily human-populated landscapes have resulted in severe ecological damage to remaining undeveloped parcels of land
as well as an abundance of human-deer conflicts in the form of property damage, deer-vehicle collisions, and increased instances of tick-borne illnesses. Among many logistical challenges associated with suburban hunting is the concern regarding long-term availability of devoted hunters. If deer density goals are to be met, managers must depend on individuals with an interest in suburban hunting who are motivated to reduce deer populations to desired levels, or at a minimum, slow their continued growth. If regulated hunting is to be used as the primary management tool for addressing overabundant deer populations in eastern Massachusetts, it is of significant importance that there is a dependable suburban hunting constituency. Without hunters willing to participate in suburban deer management programs annually, it is unlikely that deer densities will ever be reduced to numbers that minimize ecological degradation and human-deer conflicts.
In 2015 a controlled deer management program designed to address the ecological impacts caused by deer was designed for the Blue Hills Reservation near Boston, Massachusetts. This development provided an opportunity to investigate the motivations, expectations, and values of suburban deer hunters through a questionnaire. This research effort consisted of a comprehensive web-based questionnaire designed to investigate suburban deer hunters’ values, motivations, perceptions, and expectations for their hunting experiences. Results of this study
will facilitate the first steps toward developing effective and lasting deer management programs in suburban eastern Massachusetts towns by using the adaptive impact management approach. MassWildlife biologists and managers will be better able to understand what motivates hunters to participate in controlled deer hunts so that they may guide communities in developing more successful, lasting deer management programs. Without devoted hunters who are willing to contribute high levels of effort to reduce deer densities, suburban deer management programs
using regulated hunting will not likely prove successful. If community members, town representatives, and state wildlife authorities communicate clearly to make realistic, science-based decisions, effective suburban deer management programs using regulated hunting may be attainable in eastern Massachusetts.Master of Science (M.S.