Matching Items (982)
Filtering by
- All Subjects: Sustainability
- All Subjects: Machine Learning
- Member of: Theses and Dissertations
- Member of: Barrett, The Honors College Thesis/Creative Project Collection
- Resource Type: Text
Description
Urban water systems face sustainability challenges ranging from water quality, leaks, over-use, energy consumption, and long-term supply concerns. Resiliency challenges include the capacity to respond to drought, managing pipe deterioration, responding to natural disasters, and preventing terrorism. One strategy to enhance sustainability and resiliency is the development and adoption of smart water grids. A smart water grid incorporates networked monitoring and control devices into its structure, which provides diverse, real-time information about the system, as well as enhanced control. Data provide input for modeling and analysis, which informs control decisions, allowing for improvement in sustainability and resiliency. While smart water grids hold much potential, there are also potential tradeoffs and adoption challenges. More publicly available cost-benefit analyses are needed, as well as system-level research and application, rather than the current focus on individual technologies. This thesis seeks to fill one of these gaps by analyzing the cost and environmental benefits of smart irrigation controllers. Smart irrigation controllers can save water by adapting watering schedules to climate and soil conditions. The potential benefit of smart irrigation controllers is particularly high in southwestern U.S. states, where the arid climate makes water scarcer and increases watering needs of landscapes. To inform the technology development process, a design for environment (DfE) method was developed, which overlays economic and environmental performance parameters under different operating conditions. This method is applied to characterize design goals for controller price and water savings that smart irrigation controllers must meet to yield life cycle carbon dioxide reductions and economic savings in southwestern U.S. states, accounting for regional variability in electricity and water prices and carbon overhead. Results from applying the model to smart irrigation controllers in the Southwest suggest that some areas are significantly easier to design for.
ContributorsMutchek, Michele (Author) / Allenby, Braden (Thesis advisor) / Williams, Eric (Committee member) / Westerhoff, Paul (Committee member) / Arizona State University (Publisher)
Created2012
Description
Image understanding has been playing an increasingly crucial role in vision applications. Sparse models form an important component in image understanding, since the statistics of natural images reveal the presence of sparse structure. Sparse methods lead to parsimonious models, in addition to being efficient for large scale learning. In sparse modeling, data is represented as a sparse linear combination of atoms from a "dictionary" matrix. This dissertation focuses on understanding different aspects of sparse learning, thereby enhancing the use of sparse methods by incorporating tools from machine learning. With the growing need to adapt models for large scale data, it is important to design dictionaries that can model the entire data space and not just the samples considered. By exploiting the relation of dictionary learning to 1-D subspace clustering, a multilevel dictionary learning algorithm is developed, and it is shown to outperform conventional sparse models in compressed recovery, and image denoising. Theoretical aspects of learning such as algorithmic stability and generalization are considered, and ensemble learning is incorporated for effective large scale learning. In addition to building strategies for efficiently implementing 1-D subspace clustering, a discriminative clustering approach is designed to estimate the unknown mixing process in blind source separation. By exploiting the non-linear relation between the image descriptors, and allowing the use of multiple features, sparse methods can be made more effective in recognition problems. The idea of multiple kernel sparse representations is developed, and algorithms for learning dictionaries in the feature space are presented. Using object recognition experiments on standard datasets it is shown that the proposed approaches outperform other sparse coding-based recognition frameworks. Furthermore, a segmentation technique based on multiple kernel sparse representations is developed, and successfully applied for automated brain tumor identification. Using sparse codes to define the relation between data samples can lead to a more robust graph embedding for unsupervised clustering. By performing discriminative embedding using sparse coding-based graphs, an algorithm for measuring the glomerular number in kidney MRI images is developed. Finally, approaches to build dictionaries for local sparse coding of image descriptors are presented, and applied to object recognition and image retrieval.
ContributorsJayaraman Thiagarajan, Jayaraman (Author) / Spanias, Andreas (Thesis advisor) / Frakes, David (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
With the increase in computing power and availability of data, there has never been a greater need to understand data and make decisions from it. Traditional statistical techniques may not be adequate to handle the size of today's data or the complexities of the information hidden within the data. Thus knowledge discovery by machine learning techniques is necessary if we want to better understand information from data. In this dissertation, we explore the topics of asymmetric loss and asymmetric data in machine learning and propose new algorithms as solutions to some of the problems in these topics. We also studied variable selection of matched data sets and proposed a solution when there is non-linearity in the matched data. The research is divided into three parts. The first part addresses the problem of asymmetric loss. A proposed asymmetric support vector machine (aSVM) is used to predict specific classes with high accuracy. aSVM was shown to produce higher precision than a regular SVM. The second part addresses asymmetric data sets where variables are only predictive for a subset of the predictor classes. Asymmetric Random Forest (ARF) was proposed to detect these kinds of variables. The third part explores variable selection for matched data sets. Matched Random Forest (MRF) was proposed to find variables that are able to distinguish case and control without the restrictions that exists in linear models. MRF detects variables that are able to distinguish case and control even in the presence of interaction and qualitative variables.
ContributorsKoh, Derek (Author) / Runger, George C. (Thesis advisor) / Wu, Tong (Committee member) / Pan, Rong (Committee member) / Cesta, John (Committee member) / Arizona State University (Publisher)
Created2013
Description
For some time it has been recognized amongst researchers that individual and collective change should be the goal in educating for sustainability, unfortunately education has generally been ineffective in developing pro-environmental behaviors among students. Still, many scholars and practitioners are counting on education to lead us towards sustainability but suggest that in order to do so we must transition away from current information-intensive education methods. In order to develop and test novel sustainability education techniques, this research integrates pedagogical methods with psychological knowledge to target well-established sustainable behaviors. Through integrating education, behavior change, and sustainability research, I aim to answer: How can we motivate sustainable behavioral change through education programs? More specifically: How do diverse knowledge domains (declarative, procedural, effectiveness, and social) influence sustainable behaviors, both in general as well as before and after a sustainability education program? And: What are barriers hindering education approaches to changing behaviors? In answering these questions, this research involved three distinct stages: (1) Developing a theoretical framework for educating for sustainability and transformative change; (2) Implementing a food and waste focused sustainability educational program with K-12 students and teachers while intensively assessing participants' change over the course of one year; (3) Developing and implementing an extensive survey that examines the quantitative relationships between diverse domains of knowledge and behavior among a large sample of K-12 educators. The results from the education program demonstrated that significant changes in knowledge and behaviors were achieved but social knowledge in terms of food was more resistant to change as compared to that of waste. The survey results demonstrated that K-12 educators have high levels of declarative (factual or technical) knowledge regarding anthropocentric impacts on the environment; however, declarative knowledge does not predict their participation in sustainable behaviors. Rather, procedural and social knowledge significantly influence participation in sustainable food behaviors, where as procedural, effectiveness, and social knowledge impact participation in sustainable waste behaviors. Overall, the findings from this research imply that in order to effectively educate for sustainability, we must move away from nature-centric approaches that focus on declarative knowledge and embrace different domains of knowledge (procedural, effectiveness, and social) that emphasis the social implications of change.
ContributorsRedman, Erin (Author) / Larson, Kelli (Thesis advisor) / Eakin, Hallie (Committee member) / Spielmann, Katherine (Committee member) / Arizona State University (Publisher)
Created2013
Description
Effective modeling of high dimensional data is crucial in information processing and machine learning. Classical subspace methods have been very effective in such applications. However, over the past few decades, there has been considerable research towards the development of new modeling paradigms that go beyond subspace methods. This dissertation focuses on the study of sparse models and their interplay with modern machine learning techniques such as manifold, ensemble and graph-based methods, along with their applications in image analysis and recovery. By considering graph relations between data samples while learning sparse models, graph-embedded codes can be obtained for use in unsupervised, supervised and semi-supervised problems. Using experiments on standard datasets, it is demonstrated that the codes obtained from the proposed methods outperform several baseline algorithms. In order to facilitate sparse learning with large scale data, the paradigm of ensemble sparse coding is proposed, and different strategies for constructing weak base models are developed. Experiments with image recovery and clustering demonstrate that these ensemble models perform better when compared to conventional sparse coding frameworks. When examples from the data manifold are available, manifold constraints can be incorporated with sparse models and two approaches are proposed to combine sparse coding with manifold projection. The improved performance of the proposed techniques in comparison to sparse coding approaches is demonstrated using several image recovery experiments. In addition to these approaches, it might be required in some applications to combine multiple sparse models with different regularizations. In particular, combining an unconstrained sparse model with non-negative sparse coding is important in image analysis, and it poses several algorithmic and theoretical challenges. A convex and an efficient greedy algorithm for recovering combined representations are proposed. Theoretical guarantees on sparsity thresholds for exact recovery using these algorithms are derived and recovery performance is also demonstrated using simulations on synthetic data. Finally, the problem of non-linear compressive sensing, where the measurement process is carried out in feature space obtained using non-linear transformations, is considered. An optimized non-linear measurement system is proposed, and improvements in recovery performance are demonstrated in comparison to using random measurements as well as optimized linear measurements.
ContributorsNatesan Ramamurthy, Karthikeyan (Author) / Spanias, Andreas (Thesis advisor) / Tsakalis, Konstantinos (Committee member) / Karam, Lina (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
Over the last two decades programs and mandates to encourage and foster sustainable urban development have arisen throughout the world, as cities have emerged as key opportunity sites for sustainable development due to the compactness and localization of services and resources. In order to recognize this potential, scholars and practitioners have turned to the practice of visioning as a way to motivate actions and decision making toward a sustainable future. A "vision" is defined as desirable state in the future and scholars believe that the creation of a shared, motivational vision is the best starting point to catalyze positive and sustainable change. However, recent studies on city visions indicate that they do not offer substantive sustainability content, and methods or processes to evaluate the sustainability content of the resulting vision (sustainability appraisal or assessment) are often absent from the visioning process. Thus, this paper explores methods for sustainability appraisal and their potential contributions to (and in) visioning. The goal is to uncover the elements of a robust sustainability appraisal and integrate them into the visioning process. I propose an integrated sustainability appraisal procedure based on sustainability criteria, indicators, and targets as part of a visioning methodology that was developed by a team of researchers at Arizona State University (ASU) of which I was a part. I demonstrate the applicability of the appraisal method in a case study of visioning in Phoenix, Arizona. The proposed method allows for early and frequent consideration and evaluation of sustainability objectives for urban development throughout the visioning process and will result in more sustainability-oriented visions. Further, it can allow for better measurement and monitoring of progress towards sustainability goals, which can make the goals more tangible and lead to more accountability for making progress towards the development of more sustainable cities in the future.
ContributorsMinowitz, Amy (Author) / Wiek, Arnim (Thesis advisor) / Golub, Aaron (Committee member) / Pfeiffer, Deirdre (Committee member) / Arizona State University (Publisher)
Created2013
Description
Sustainability visioning (i.e. the construction of sustainable future states) is considered an important component of sustainability research, for instance, in transformational sustainability science or in planning for urban sustainability. Visioning frees sustainability research from the dominant focus on analyzing problem constellations and opens it towards positive contributions to social innovation and transformation. Calls are repeatedly made for visions that can guide us towards sustainable futures. Scattered across a broad range of fields (i.e. business, non-government organization, land-use management, natural resource management, sustainability science, urban and regional planning) are an abundance of visioning studies. However, among the few evaluative studies in the literature there are apparent deficits in both the research and practice of visioning that curtails our expectations and prospects of realizing process-based and product-derived outcomes. These deficits suggests that calls instead should focus on the development of applied and theoretical understanding of crafting sustainability visions, enhancing the rigor and robustness of visioning methodology, and on integrating practice, research, and education for collaborative sustainability visioning. From an analysis of prominent visioning and sustainability visioning studies in the literature, this dissertation articulates what is sustainability visioning and synthesizes a conceptual framework for criteria-based design and evaluation of sustainability visioning studies. While current visioning methodologies comply with some of these guidelines, none adhere to all of them. From this research, a novel sustainability visioning methodology is designed to address this gap to craft visions that are shared, systemic, principles-based, action-oriented, relevant, and creative (i.e. SPARC visioning methodology) and evaluated across all quality criteria. Empirical studies were conducted to test and apply the conceptual and methodological frameworks -- with an emphasis on enhancing the rigor and robustness in real world visioning processes for urban planning and teaching sustainability competencies. In-depth descriptions of the collaborative visioning studies demonstrate tangible outcomes for: (a) implementing the above sustainability visioning methodology, including evaluative procedures; (b) adopting meaningful interactive engagement procedures; (c) integrating advanced analytical modeling, sustainability appraisal, and creativity enhancing procedures; and (d) developing perspective and methodological capacity for long-range sustainability planning.
ContributorsIwaniec, David (Author) / Wiek, Arnim (Thesis advisor) / Childers, Daniel L. (Committee member) / Lant, Timothy (Committee member) / Arizona State University (Publisher)
Created2013
Description
A methodology is developed that integrates institutional analysis with Life Cycle Assessment (LCA) to identify and overcome barriers to sustainability transitions and to bridge the gap between environmental practitioners and decisionmakers. LCA results are rarely joined with analyses of the social systems that control or influence decisionmaking and policies. As a result, LCA conclusions generally lack information about who or what controls different parts of the system, where and when the processes' environmental decisionmaking happens, and what aspects of the system (i.e. a policy or regulatory requirement) would have to change to enable lower environmental impact futures. The value of the combined institutional analysis and LCA (the IA-LCA) is demonstrated using a case study of passenger transportation in the Phoenix, Arizona metropolitan area. A retrospective LCA is developed to estimate how roadway investment has enabled personal vehicle travel and its associated energy, environmental, and economic effects. Using regional travel forecasts, a prospective life cycle inventory is developed. Alternative trajectories are modeled to reveal future "savings" from reduced roadway construction and vehicle travel. An institutional analysis matches the LCA results with the specific institutions, players, and policies that should be targeted to enable transitions to these alternative futures. The results show that energy, economic, and environmental benefits from changes in passenger transportation systems are possible, but vary significantly depending on the timing of the interventions. Transition strategies aimed at the most optimistic benefits should include 1) significant land-use planning initiatives at the local and regional level to incentivize transit-oriented development infill and urban densification, 2) changes to state or federal gasoline taxes, 3) enacting a price on carbon, and 4) nearly doubling vehicle fuel efficiency together with greater market penetration of alternative fuel vehicles. This aggressive trajectory could decrease the 2050 energy consumption to 1995 levels, greenhouse gas emissions to 1995, particulate emissions to 2006, and smog-forming emissions to 1972. The potential benefits and costs are both private and public, and the results vary when transition strategies are applied in different spatial and temporal patterns.
ContributorsKimball, Mindy (Author) / Chester, Mikhail (Thesis advisor) / Allenby, Braden (Committee member) / Golub, Aaron (Committee member) / Arizona State University (Publisher)
Created2014
Description
ABSTRACT Recent studies indicate that top-performing companies have higher-performing work environments than average companies. They receive higher scores for worker satisfaction with their overall physical work environment as well as higher effectiveness ratings for their workspaces (Gensler, 2008; Harter et al., 2003). While these studies indicate a relationship between effective office design and satisfaction they have not explored which specific space types may contribute to workers' overall satisfaction with their physical work environment. Therefore, the purpose of this study is to explore the relationship between workers' overall satisfaction with their physical work environments and their perception of the effectiveness of spaces designed for Conceptual Age work including learning, focusing, collaborating, and socializing tasks. This research is designed to identify which workspace types are related to workers' satisfaction with their overall work environment and which are perceived to be most and least effective. To accomplish this two primary and four secondary research questions were developed for this study. The first primary question considers overall workers' satisfaction with their overall physical work environments (offices, workstations, hallways, common areas, reception, waiting areas, etc.) related to the effective use of work mode workspaces (learning, focusing, collaborating, socializing). The second primary research question was developed to identify which of the four work mode space types had the greatest and least relationship to workers' satisfaction with the overall physical work environment. Secondary research questions were developed to address workers' perceptions of effectiveness of each space type. This research project used data from a previous study collected from 2007 to 2012. Responses were from all staff levels of US office-based office workers and resulted in a blind sample of approximately 48,000 respondents. The data for this study were developed from SPSS data reports that included descriptive data and Pearson correlations. Findings were developed from those statistics using coefficient of determination.
ContributorsHarmon-Vaughan, Elizabeth (Author) / Kroelinger, Michael D. (Thesis advisor) / Bernardi, Jose (Committee member) / Ozel, Filiz (Committee member) / Arizona State University (Publisher)
Created2013
Description
Extreme hot-weather events have become life-threatening natural phenomena in many cities around the world, and the health impacts of excessive heat are expected to increase with climate change (Huang et al. 2011; Knowlton et al. 2007; Meehl and Tebaldi 2004; Patz 2005). Heat waves will likely have the worst health impacts in urban areas, where large numbers of vulnerable people reside and where local-scale urban heat island effects (UHI) retard and reduce nighttime cooling. This dissertation presents three empirical case studies that were conducted to advance our understanding of human vulnerability to heat in coupled human-natural systems. Using vulnerability theory as a framework, I analyzed how various social and environmental components of a system interact to exacerbate or mitigate heat impacts on human health, with the goal of contributing to the conceptualization of human vulnerability to heat. The studies: 1) compared the relationship between temperature and health outcomes in Chicago and Phoenix; 2) compared a map derived from a theoretical generic index of vulnerability to heat with a map derived from actual heat-related hospitalizations in Phoenix; and 3) used geospatial information on health data at two areal units to identify the hot spots for two heat health outcomes in Phoenix. The results show a 10-degree Celsius difference in the threshold temperatures at which heat-stress calls in Phoenix and Chicago are likely to increase drastically, and that Chicago is likely to be more sensitive to climate change than Phoenix. I also found that heat-vulnerability indices are sensitive to scale, measurement, and context, and that cities will need to incorporate place-based factors to increase the usefulness of vulnerability indices and mapping to decision making. Finally, I found that identification of geographical hot-spot of heat-related illness depends on the type of data used, scale of measurement, and normalization procedures. I recommend using multiple datasets and different approaches to spatial analysis to overcome this limitation and help decision makers develop effective intervention strategies.
ContributorsChuang, Wen-Ching (Author) / Gober, Patricia (Thesis advisor) / Boone, Christopher (Committee member) / Guhathakurta, Subhrajit (Committee member) / Ruddell, Darren (Committee member) / Arizona State University (Publisher)
Created2013