- home
- Search
Filters
Clear All- Rural Digital Europe
- Doctoral thesis
- National Science Foundation
- Rural Digital Europe
- Doctoral thesis
- National Science Foundation
Loading
description Publicationkeyboard_double_arrow_right Doctoral thesis 2017 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | CyberSEES: Type 1: Foster...NSF| CyberSEES: Type 1: Fostering Non-Expert Creation of Sustainable Polycultures through Crowdsourced Data SynthesisAuthors: Raturi, Ankita;Raturi, Ankita;Agriculture is a critical component of the human food system. Its coupling to the success of human societies and its impact on the environment is nontrivial. Varied efforts -- including new regulations, certifications, techniques and software -- exist to assess and improve the sustainability of agriculture. Multiple stakeholders in a fragmented field, with tensions and pulls in different directions, results in a duplication of efforts and disconnected data and processes.To explore the challenges that exist in modeling sustainable agriculture, I characterize environmental assessment as a modeling process, and secondly, characterize sustainable agricultural systems as a type of complex adaptive system. Framing the assessment process and system of interest in this manner permits the application of various techniques from software engineering, systems analysis, and human-computer interaction to tease apart the core issues and to subsequently respond to these challenges through design.First, I present an analysis of the capacity of Life Cycle Assessment (a formal and quantitative environmental assessment technique) to represent small- to medium-scale sustainability-oriented farms. Then, I described a qualitative field study, in which I visited 16 farms across California, interviewing sustainability-oriented farmers, and collecting samples of farm data. The goal of this study was to uncover how and why farmers model farms in practice, the nature and availability of farm data, and the experiences of farmers with various environmental assessment techniques.The findings of these two studies resulted in the articulation of domain-specific modeling requirements. These include: creating selective and partial system models, knitting together qualitative and quantitative data in system models, capturing both spatial and temporal structures, and all of this through models that are abstract yet grounded in real farm data.Building on these studies, I present MoSS: a framework to enable the Modeling of Sustainable Systems. MoSS consists of three parts: an abstract model, domain-specific elements to allow for modeling agricultural systems, and model 'perspectives' that allow for the assessment of the environmental performance of the system. I conducted a scenario-based evaluation of MoSS to assess its ability to express the varying dynamism and complexity of sustainable agricultural systems. MoSS addresses the core challenges involved in modeling sustainable agriculture, providing a consistent mechanism to capture the essence of farms.MoSS represents a step forward in grounded information design for sustainable agriculture, paving the way for the design of information management and environmental assessment tools that more closely meet the needs of small- to medium-scale farms and farmers. Through the work presented in this dissertation, I have also demonstrated how one may engage in applied and interdisciplinary software engineering research to support sustainable development.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2017Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::5c4b9ceae4e0af07a9f5c046115b31d9&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2017Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::5c4b9ceae4e0af07a9f5c046115b31d9&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2023Publisher:University of Virginia Funded by:NSF | NRT: A Graduate Trainees..., NSF | Phase I IUCRC University ...NSF| NRT: A Graduate Traineeship in Cyber Physical Systems ,NSF| Phase I IUCRC University of Virginia: Center for Hardware and Embedded System Security and Trust (CHEST)Authors: William Clark;William Clark;doi: 10.18130/5xpn-s618
Decision making under uncertainty is a central problem for modern robot autonomy. As robots are deployed into the chaos of the wider world, they are required to make more complex decisions with imperfect knowledge. In order to cope with this disorder, and the resulting uncertainty, a panoply of techniques have been developed which minimize or otherwise regulate uncertainty. However, such techniques lack a consideration of consequences. It is this notion of consequences and uncertainty blended together that forms the conceptual backbone of risk analysis, and it is only recently that robotics has begun to embrace these techniques. Drawing on a rich theory of coherent risk measures, originally developed in the financial sector to aid in selecting safe investments, novel approaches to autonomous decision making have been developed that allow robots to properly evaluate risk in their decision making. In order to compute their risk measures, these approaches have generally required either highly problem specific formulations for their optimization, or that the risks be describable using well behaved functions like Gaussians. In our approach, we devise a sampling-based means of computing risk measures that admits a wide range of possible risk metrics. We develop a technique using Hamiltonian Monte Carlo to sample the stochastic reachable set of a robot, which we convert to a distribution of consequences for estimating the Conditional Value-at-Risk for a proposed action by the robot. We also develop a scheme to overcome Hamiltonian Monte Carlo's inability to sample across abrupt changes in dynamics by partitioning the problem into single-dynamic segments, and propagating uncertainty forward across the segments. Having developed a technique to estimate risk, we also propose a scheme using estimates of the risk of collision to enable safe navigation. Using a Probabilistic Roadmap and the A* algorithm, we generate a series of waypoints directing our robot to its goal. These waypoints are fed to a pure-pursuit controller, which we use to generate controls to compute the risk of collision. Based on this risk, we can accept the path, or reject it, and use a heuristic update to find a new path. We show simulations as proof of concept for this approach.
add ClaimPlease grant OpenAIRE to access and update your ORCID works.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=10.18130/5xpn-s618&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert add ClaimPlease grant OpenAIRE to access and update your ORCID works.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=10.18130/5xpn-s618&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2019 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | Collaborative Research: V...NSF| Collaborative Research: Variability-Aware Software for Efficient Computing with Nanoscale DevicesAuthors: Lin, Jeng-Hau;Lin, Jeng-Hau;The entangled guardbands in terms of timing specification and energy budget ensure a system against faults, but the guardbands, meanwhile, impede the advance of a higher throughput and energy efficiency. To combat the over-designed guardbands in a system carrying out deep learning inference, we dive into the algorithmic demands and understand that the resource deficiency and hardware variation are the major reasons of the need of conservative guardbands. In modern convolutional neural networks (CNNs), the number of arithmetic operations for the inference could exceed tens of billions, which requires a sophisticated buffering mechanism to balance between resource utilization and throughput. In this case, the over-designed guardbands can seriously hinder system performance. On the other hand, timing errors can be incurred by the hardware variations including momentary voltage droops resulted from simultaneous switching noises, a gradually decreasing voltage level due to a limited battery, and the slow electron mobility incurred by the system power dissipation into heat. The timing errors propagating in a network can be a snowball in the beginning but ends up with a catastrophe in terms of a significant accuracy degradation.Knowing the need of guardbands originates from resource deficiency and timing errors, this dissertation focuses on cross-layer solutions to the problems of the high algorithmic demands incurred by deep learning methods and error vulnerability due to hardware variations. We begin with reviewing the methods and technologies proposed in the literature including weight encoding, filter decomposition, network pruning, efficient structure design, and precision quantizing. In the implementation of an FPGA accelerator for extreme-case quantization, binarized neural networks (BNN), we have realized more possible optimizations can be applied. Then, we extend BNN on the algorithmic layer with the binarized separable filters and proposed BCNNw/SF. Although the quantization and approximation benefit hardware efficiency to a certain extent, the optimal reduction or compression rate is still limited by the core of the conventional deep learning methods -- convolution. We, thus, introduce the local binary pattern (LBP) to deep learning because of LBP's low complexity yet high effectiveness. We name the new algorithm LBPNet, in which the feature maps are created with a similar fashion of the traditional LBP using comparisons. Our LBPNet can be trained with the forward-backward propagation algorithm to extract useful features for image classification. LBPNet accelerators have been implemented and optimized to verify their classification performance, processing throughput, and energy efficiency. We also demonstrate the error immunity of LBPNet to be the strongest compared with the subject MLP, CNN, and BCNN models since the classification accuracy of the LBPNet is decreased by only 10% and all the other models lose the classification ability when the timing error rate exceeds 0.01.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2019Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::6078c60fd0cdac7856891bf72ba6ff22&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2019Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::6078c60fd0cdac7856891bf72ba6ff22&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2012 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | US-Based Students Support..., NSF | III:Small:RUI:Integrating...NSF| US-Based Students Support to Attend the ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems 2010 (ACM SIGSPATIAL GIS 2010) ,NSF| III:Small:RUI:Integrating Image and Non-Image Geospatial DataAuthors: Yang, Yang;Yang, Yang;High resolution remote sensed image data continues to become more accessible. One consequence of this is that novel geographic information system are playing an increasingly important role not only for academia, but also for daily human business and life. Nevertheless, to automate the understanding of the exponentially growing geographic image data repositories remains by-and-large an unsolved problem. In this dissertation, we put forward efforts to tackle the most important and comprehensive problems in understanding the remotely sensed image data: image retrieval, classification and object recognition. In the interest of high resolution overhead images, we adapt and extend techniques that have been highly developed in generic vision tasks. We investigate the applications of low-level local descriptors to the remote sensed image analysis. In particular, we evaluate how local invariant descriptors perform compared to proven global texture as well as color features for similarity retrieval. We further investigate how different similarity measurements and sizes of the set of interest points used to represent images influence the retrieval. In addition, we extend our work to image classification using bagof- visual-words models. Moreover, we explore the potential for increased synergy between two complementary data sources: gazetteers and overhead imagery. We explore ways in which these two data sources can be integrated to more fully automate geographic data management. In particular, we propose a hieararchial model to estimate the spatial extents of archived geospatial objects from gazetteers such that their spatial representations can be extended from a single latitude/longtidue pair to a bounding box.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2012Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::8aa06c67bcc92d19b741053f8f464fc6&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2012Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::8aa06c67bcc92d19b741053f8f464fc6&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2017 France FrenchPublisher:HAL CCSD Funded by:NSERC, NSF | WSC Category-2: Climate C..., NSF | LTER: The Interacting Eff... +4 projectsNSERC ,NSF| WSC Category-2: Climate Change, Shifting Land Use, and Urbanization in a Midwestern Agricultural Landscape: Challenges for Water Quality and Quantity ,NSF| LTER: The Interacting Effects of Hydroclimate Variability and Human Landscape Modification in the Southern Appalachian Mountains ,EC| OPENNESS ,NSF| LTER: Comparative Study of a Suite of Lakes in Wisconsin ,EC| GLOLAND ,EC| OPERASAuthors: Lasseur, Rémy;Lasseur, Rémy;Societal changes over the last century have strongly affected the majority of ecosystem dynamics. Our society is facing a complex paradox: how to maintain decent livelihoods for the world population while limiting negative effects of human activities on ecosystems? The concept of ecosystem services has been proposed to contribute to the solving of this paradox and it holds strong expectations. My PhD research aims at assessing the contribution of ecosystem services mapping to the definition of patterns and drivers of landscape multifunctionality.Chapter I analyses the spatial associations between the agricultural production service and four regulating services over the agricultural lands of the Rhône-Alpes region (France). This work allows us to insist on the ways trade-off and synergies between ecosystem services are defined as well as on the characterization of bundles of ecosystem services. Furthermore, taking advantage of the "ecological niches” concept usually applied to biodiversity, we assess spatial matching between the supply of ecosystem services and the socio-ecological specificities of associated areas, i.e. their social-ecological niche. Our results highlight a large variability concerning associations between agricultural production and regulating services, which illustrates the strong influence of farming practices (e.g., in terms of intensity) in defining the strength of associations between multiple services. In addition, we raised the issue of the robustness of standard statistical analyses to consistently identify bundles of ecosystem services.Chapter II assesses the influence of spatial resolution of modeled data on ecosystem services mapping. To this end, we compare the maps used in the first chapter with high spatial resolution data provided at Grenoble area scale (in the context of the ESNET project). Based on this comparison, we discuss the limits of models used to map ecosystem services.To improve the mapping of ecosystem services supplied by agricultural areas, we propose, in chapter III, a remote sensing-based approach to map agricultural land uses at high resolution on Grenoble region. Simultaneous use of MODIS and Rapideye satellite data allows us to determine cropping successions for 5 years at farming plot scale. These spatially explicit data significantly improved our abilities to map agricultural productions and may be used to map several other ecosystem services.To complete the third chapter, chapter IV gives a synthesis of remote sensing approaches that could be used to map ecosystem services, focusing on methods that are not linked to land uses identification. Based on a wide panel of ecosystem services mapping studies, we highlight data currently needed to map ecosystem services. Then we bridge these needs and the potential of remote sensing approaches for ecosystem services modelers.; Les mutations de la société au cours du dernier siècle ont de fortes répercussions sur le fonctionnement de la majorité des écosystèmes. Notre société fait face à un paradoxe complexe défini par la nécessité de satisfaire des conditions de vie décentes d’une population mondiale croissante tout en limitant les impacts négatifs sur les écosystèmes. Le concept de service écosystémique a été proposé pour accompagner la résolution de ce paradoxe et de nombreuses attentes reposent sur ce concept. Mon travail de thèse a pour objectifs d’évaluer les apports actuels de la cartographie des services écosystémiques pour caractériser les patrons spatiaux et déterminants de la multifonctionnalité des territoires.Dans le chapitre I nous proposons une analyse des associations entre le service de production agricole et quatre services de régulation au niveau des surfaces agricoles de la région Rhône-Alpes (France). Ce travail nous permet de revenir sur la définition des compromis et synergies entre services écosystémiques aussi bien que sur le concept de bouquet de services écosystèmiques. Par ailleurs, en s’inspirant du concept de niche écologique pour les espèces, nous évaluons l’adéquation spatiale entre la fourniture de services écosystèmiques et les caractéristiques socio-écologiques des milieux (i.e. leur niche socio-écologique). Nos résultats révèlent une grande variabilité dans les associations entre le service de production agricole et les services de régulation. Cette observation souligne le poids des modalités de gestion agricole dans l’orientation des associations entre services écosystèmiques. Par ailleurs, nous soulevons de nombreuses interrogations vis-à-vis de la robustesse des analyses courantes des bouquets de services écosystèmiques.L’influence de la résolution spatiale des données de modélisation sur la cartographie des services écosystèmiques est évaluée dans le chapitre II. Pour cela nous comparons les cartes utilisées dans le premier chapitre avec les informations apportées par un projet de cartographie de services écosystèmiques à haute résolution spatiale sur la région de Grenoble (projet ESNET). Cette analyse alimente une réflexion sur les limites des modèles utilisés pour cartographier les services écosystémiques.Dans le but d’améliorer la modélisation des services écosystémiques fournis par les terres agricoles, nous développons dans le chapitre III, une approche basée sur la télédétection pour cartographier l’utilisation des terres agricoles dans la région de Grenoble. L’utilisation conjointe de données satellitaires MODIS et RapidEye nous permet de déterminer les successions culturales sur 5 années à la résolution de la parcelle agricole. Validée par les données du registre parcellaire graphique, l’utilisation de ces données spatialement explicites améliore significativement notre capacité de cartographie de la production agricole et pourrait être utilisée pour cartographier de nombreux autres services écosystèmiques.Alors que le chapitre III est focalisé sur l’utilisation de la télédétection pour la cartographie de l’utilisation des terres agricoles, le chapitre IV propose une synthèse des autres utilisations des données de télédétection pouvant potentiellement contribuer à la modélisation des services écosystémiques. Ce travail est une mise en correspondance entre les capacités des approches de télédétection et les besoins des modélisateurs de services écosystémiques, sur la base d’une analyse bibliographique d’un large panel d’études de cartographie des services écosystémiques
Hal-Diderot arrow_drop_down Mémoires en Sciences de l'Information et de la CommunicationDoctoral thesis . 2017All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______212::3f29e330e5fde3148e3f881329d8ee4c&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert Hal-Diderot arrow_drop_down Mémoires en Sciences de l'Information et de la CommunicationDoctoral thesis . 2017All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______212::3f29e330e5fde3148e3f881329d8ee4c&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu
Loading
description Publicationkeyboard_double_arrow_right Doctoral thesis 2017 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | CyberSEES: Type 1: Foster...NSF| CyberSEES: Type 1: Fostering Non-Expert Creation of Sustainable Polycultures through Crowdsourced Data SynthesisAuthors: Raturi, Ankita;Raturi, Ankita;Agriculture is a critical component of the human food system. Its coupling to the success of human societies and its impact on the environment is nontrivial. Varied efforts -- including new regulations, certifications, techniques and software -- exist to assess and improve the sustainability of agriculture. Multiple stakeholders in a fragmented field, with tensions and pulls in different directions, results in a duplication of efforts and disconnected data and processes.To explore the challenges that exist in modeling sustainable agriculture, I characterize environmental assessment as a modeling process, and secondly, characterize sustainable agricultural systems as a type of complex adaptive system. Framing the assessment process and system of interest in this manner permits the application of various techniques from software engineering, systems analysis, and human-computer interaction to tease apart the core issues and to subsequently respond to these challenges through design.First, I present an analysis of the capacity of Life Cycle Assessment (a formal and quantitative environmental assessment technique) to represent small- to medium-scale sustainability-oriented farms. Then, I described a qualitative field study, in which I visited 16 farms across California, interviewing sustainability-oriented farmers, and collecting samples of farm data. The goal of this study was to uncover how and why farmers model farms in practice, the nature and availability of farm data, and the experiences of farmers with various environmental assessment techniques.The findings of these two studies resulted in the articulation of domain-specific modeling requirements. These include: creating selective and partial system models, knitting together qualitative and quantitative data in system models, capturing both spatial and temporal structures, and all of this through models that are abstract yet grounded in real farm data.Building on these studies, I present MoSS: a framework to enable the Modeling of Sustainable Systems. MoSS consists of three parts: an abstract model, domain-specific elements to allow for modeling agricultural systems, and model 'perspectives' that allow for the assessment of the environmental performance of the system. I conducted a scenario-based evaluation of MoSS to assess its ability to express the varying dynamism and complexity of sustainable agricultural systems. MoSS addresses the core challenges involved in modeling sustainable agriculture, providing a consistent mechanism to capture the essence of farms.MoSS represents a step forward in grounded information design for sustainable agriculture, paving the way for the design of information management and environmental assessment tools that more closely meet the needs of small- to medium-scale farms and farmers. Through the work presented in this dissertation, I have also demonstrated how one may engage in applied and interdisciplinary software engineering research to support sustainable development.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2017Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::5c4b9ceae4e0af07a9f5c046115b31d9&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2017Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::5c4b9ceae4e0af07a9f5c046115b31d9&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2023Publisher:University of Virginia Funded by:NSF | NRT: A Graduate Trainees..., NSF | Phase I IUCRC University ...NSF| NRT: A Graduate Traineeship in Cyber Physical Systems ,NSF| Phase I IUCRC University of Virginia: Center for Hardware and Embedded System Security and Trust (CHEST)Authors: William Clark;William Clark;doi: 10.18130/5xpn-s618
Decision making under uncertainty is a central problem for modern robot autonomy. As robots are deployed into the chaos of the wider world, they are required to make more complex decisions with imperfect knowledge. In order to cope with this disorder, and the resulting uncertainty, a panoply of techniques have been developed which minimize or otherwise regulate uncertainty. However, such techniques lack a consideration of consequences. It is this notion of consequences and uncertainty blended together that forms the conceptual backbone of risk analysis, and it is only recently that robotics has begun to embrace these techniques. Drawing on a rich theory of coherent risk measures, originally developed in the financial sector to aid in selecting safe investments, novel approaches to autonomous decision making have been developed that allow robots to properly evaluate risk in their decision making. In order to compute their risk measures, these approaches have generally required either highly problem specific formulations for their optimization, or that the risks be describable using well behaved functions like Gaussians. In our approach, we devise a sampling-based means of computing risk measures that admits a wide range of possible risk metrics. We develop a technique using Hamiltonian Monte Carlo to sample the stochastic reachable set of a robot, which we convert to a distribution of consequences for estimating the Conditional Value-at-Risk for a proposed action by the robot. We also develop a scheme to overcome Hamiltonian Monte Carlo's inability to sample across abrupt changes in dynamics by partitioning the problem into single-dynamic segments, and propagating uncertainty forward across the segments. Having developed a technique to estimate risk, we also propose a scheme using estimates of the risk of collision to enable safe navigation. Using a Probabilistic Roadmap and the A* algorithm, we generate a series of waypoints directing our robot to its goal. These waypoints are fed to a pure-pursuit controller, which we use to generate controls to compute the risk of collision. Based on this risk, we can accept the path, or reject it, and use a heuristic update to find a new path. We show simulations as proof of concept for this approach.
add ClaimPlease grant OpenAIRE to access and update your ORCID works.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=10.18130/5xpn-s618&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert add ClaimPlease grant OpenAIRE to access and update your ORCID works.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.This Research product is the result of merged Research products in OpenAIRE.
You have already added works in your ORCID record related to the merged Research product.All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=10.18130/5xpn-s618&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2019 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | Collaborative Research: V...NSF| Collaborative Research: Variability-Aware Software for Efficient Computing with Nanoscale DevicesAuthors: Lin, Jeng-Hau;Lin, Jeng-Hau;The entangled guardbands in terms of timing specification and energy budget ensure a system against faults, but the guardbands, meanwhile, impede the advance of a higher throughput and energy efficiency. To combat the over-designed guardbands in a system carrying out deep learning inference, we dive into the algorithmic demands and understand that the resource deficiency and hardware variation are the major reasons of the need of conservative guardbands. In modern convolutional neural networks (CNNs), the number of arithmetic operations for the inference could exceed tens of billions, which requires a sophisticated buffering mechanism to balance between resource utilization and throughput. In this case, the over-designed guardbands can seriously hinder system performance. On the other hand, timing errors can be incurred by the hardware variations including momentary voltage droops resulted from simultaneous switching noises, a gradually decreasing voltage level due to a limited battery, and the slow electron mobility incurred by the system power dissipation into heat. The timing errors propagating in a network can be a snowball in the beginning but ends up with a catastrophe in terms of a significant accuracy degradation.Knowing the need of guardbands originates from resource deficiency and timing errors, this dissertation focuses on cross-layer solutions to the problems of the high algorithmic demands incurred by deep learning methods and error vulnerability due to hardware variations. We begin with reviewing the methods and technologies proposed in the literature including weight encoding, filter decomposition, network pruning, efficient structure design, and precision quantizing. In the implementation of an FPGA accelerator for extreme-case quantization, binarized neural networks (BNN), we have realized more possible optimizations can be applied. Then, we extend BNN on the algorithmic layer with the binarized separable filters and proposed BCNNw/SF. Although the quantization and approximation benefit hardware efficiency to a certain extent, the optimal reduction or compression rate is still limited by the core of the conventional deep learning methods -- convolution. We, thus, introduce the local binary pattern (LBP) to deep learning because of LBP's low complexity yet high effectiveness. We name the new algorithm LBPNet, in which the feature maps are created with a similar fashion of the traditional LBP using comparisons. Our LBPNet can be trained with the forward-backward propagation algorithm to extract useful features for image classification. LBPNet accelerators have been implemented and optimized to verify their classification performance, processing throughput, and energy efficiency. We also demonstrate the error immunity of LBPNet to be the strongest compared with the subject MLP, CNN, and BCNN models since the classification accuracy of the LBPNet is decreased by only 10% and all the other models lose the classification ability when the timing error rate exceeds 0.01.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2019Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::6078c60fd0cdac7856891bf72ba6ff22&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2019Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::6078c60fd0cdac7856891bf72ba6ff22&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2012 United States EnglishPublisher:eScholarship, University of California Funded by:NSF | US-Based Students Support..., NSF | III:Small:RUI:Integrating...NSF| US-Based Students Support to Attend the ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems 2010 (ACM SIGSPATIAL GIS 2010) ,NSF| III:Small:RUI:Integrating Image and Non-Image Geospatial DataAuthors: Yang, Yang;Yang, Yang;High resolution remote sensed image data continues to become more accessible. One consequence of this is that novel geographic information system are playing an increasingly important role not only for academia, but also for daily human business and life. Nevertheless, to automate the understanding of the exponentially growing geographic image data repositories remains by-and-large an unsolved problem. In this dissertation, we put forward efforts to tackle the most important and comprehensive problems in understanding the remotely sensed image data: image retrieval, classification and object recognition. In the interest of high resolution overhead images, we adapt and extend techniques that have been highly developed in generic vision tasks. We investigate the applications of low-level local descriptors to the remote sensed image analysis. In particular, we evaluate how local invariant descriptors perform compared to proven global texture as well as color features for similarity retrieval. We further investigate how different similarity measurements and sizes of the set of interest points used to represent images influence the retrieval. In addition, we extend our work to image classification using bagof- visual-words models. Moreover, we explore the potential for increased synergy between two complementary data sources: gazetteers and overhead imagery. We explore ways in which these two data sources can be integrated to more fully automate geographic data management. In particular, we propose a hieararchial model to estimate the spatial extents of archived geospatial objects from gazetteers such that their spatial representations can be extended from a single latitude/longtidue pair to a bounding box.
eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2012Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::8aa06c67bcc92d19b741053f8f464fc6&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert eScholarship - Unive... arrow_drop_down eScholarship - University of CaliforniaDoctoral thesis . 2012Data sources: eScholarship - University of CaliforniaAll Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______325::8aa06c67bcc92d19b741053f8f464fc6&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eudescription Publicationkeyboard_double_arrow_right Doctoral thesis 2017 France FrenchPublisher:HAL CCSD Funded by:NSERC, NSF | WSC Category-2: Climate C..., NSF | LTER: The Interacting Eff... +4 projectsNSERC ,NSF| WSC Category-2: Climate Change, Shifting Land Use, and Urbanization in a Midwestern Agricultural Landscape: Challenges for Water Quality and Quantity ,NSF| LTER: The Interacting Effects of Hydroclimate Variability and Human Landscape Modification in the Southern Appalachian Mountains ,EC| OPENNESS ,NSF| LTER: Comparative Study of a Suite of Lakes in Wisconsin ,EC| GLOLAND ,EC| OPERASAuthors: Lasseur, Rémy;Lasseur, Rémy;Societal changes over the last century have strongly affected the majority of ecosystem dynamics. Our society is facing a complex paradox: how to maintain decent livelihoods for the world population while limiting negative effects of human activities on ecosystems? The concept of ecosystem services has been proposed to contribute to the solving of this paradox and it holds strong expectations. My PhD research aims at assessing the contribution of ecosystem services mapping to the definition of patterns and drivers of landscape multifunctionality.Chapter I analyses the spatial associations between the agricultural production service and four regulating services over the agricultural lands of the Rhône-Alpes region (France). This work allows us to insist on the ways trade-off and synergies between ecosystem services are defined as well as on the characterization of bundles of ecosystem services. Furthermore, taking advantage of the "ecological niches” concept usually applied to biodiversity, we assess spatial matching between the supply of ecosystem services and the socio-ecological specificities of associated areas, i.e. their social-ecological niche. Our results highlight a large variability concerning associations between agricultural production and regulating services, which illustrates the strong influence of farming practices (e.g., in terms of intensity) in defining the strength of associations between multiple services. In addition, we raised the issue of the robustness of standard statistical analyses to consistently identify bundles of ecosystem services.Chapter II assesses the influence of spatial resolution of modeled data on ecosystem services mapping. To this end, we compare the maps used in the first chapter with high spatial resolution data provided at Grenoble area scale (in the context of the ESNET project). Based on this comparison, we discuss the limits of models used to map ecosystem services.To improve the mapping of ecosystem services supplied by agricultural areas, we propose, in chapter III, a remote sensing-based approach to map agricultural land uses at high resolution on Grenoble region. Simultaneous use of MODIS and Rapideye satellite data allows us to determine cropping successions for 5 years at farming plot scale. These spatially explicit data significantly improved our abilities to map agricultural productions and may be used to map several other ecosystem services.To complete the third chapter, chapter IV gives a synthesis of remote sensing approaches that could be used to map ecosystem services, focusing on methods that are not linked to land uses identification. Based on a wide panel of ecosystem services mapping studies, we highlight data currently needed to map ecosystem services. Then we bridge these needs and the potential of remote sensing approaches for ecosystem services modelers.; Les mutations de la société au cours du dernier siècle ont de fortes répercussions sur le fonctionnement de la majorité des écosystèmes. Notre société fait face à un paradoxe complexe défini par la nécessité de satisfaire des conditions de vie décentes d’une population mondiale croissante tout en limitant les impacts négatifs sur les écosystèmes. Le concept de service écosystémique a été proposé pour accompagner la résolution de ce paradoxe et de nombreuses attentes reposent sur ce concept. Mon travail de thèse a pour objectifs d’évaluer les apports actuels de la cartographie des services écosystémiques pour caractériser les patrons spatiaux et déterminants de la multifonctionnalité des territoires.Dans le chapitre I nous proposons une analyse des associations entre le service de production agricole et quatre services de régulation au niveau des surfaces agricoles de la région Rhône-Alpes (France). Ce travail nous permet de revenir sur la définition des compromis et synergies entre services écosystémiques aussi bien que sur le concept de bouquet de services écosystèmiques. Par ailleurs, en s’inspirant du concept de niche écologique pour les espèces, nous évaluons l’adéquation spatiale entre la fourniture de services écosystèmiques et les caractéristiques socio-écologiques des milieux (i.e. leur niche socio-écologique). Nos résultats révèlent une grande variabilité dans les associations entre le service de production agricole et les services de régulation. Cette observation souligne le poids des modalités de gestion agricole dans l’orientation des associations entre services écosystèmiques. Par ailleurs, nous soulevons de nombreuses interrogations vis-à-vis de la robustesse des analyses courantes des bouquets de services écosystèmiques.L’influence de la résolution spatiale des données de modélisation sur la cartographie des services écosystèmiques est évaluée dans le chapitre II. Pour cela nous comparons les cartes utilisées dans le premier chapitre avec les informations apportées par un projet de cartographie de services écosystèmiques à haute résolution spatiale sur la région de Grenoble (projet ESNET). Cette analyse alimente une réflexion sur les limites des modèles utilisés pour cartographier les services écosystémiques.Dans le but d’améliorer la modélisation des services écosystémiques fournis par les terres agricoles, nous développons dans le chapitre III, une approche basée sur la télédétection pour cartographier l’utilisation des terres agricoles dans la région de Grenoble. L’utilisation conjointe de données satellitaires MODIS et RapidEye nous permet de déterminer les successions culturales sur 5 années à la résolution de la parcelle agricole. Validée par les données du registre parcellaire graphique, l’utilisation de ces données spatialement explicites améliore significativement notre capacité de cartographie de la production agricole et pourrait être utilisée pour cartographier de nombreux autres services écosystèmiques.Alors que le chapitre III est focalisé sur l’utilisation de la télédétection pour la cartographie de l’utilisation des terres agricoles, le chapitre IV propose une synthèse des autres utilisations des données de télédétection pouvant potentiellement contribuer à la modélisation des services écosystémiques. Ce travail est une mise en correspondance entre les capacités des approches de télédétection et les besoins des modélisateurs de services écosystémiques, sur la base d’une analyse bibliographique d’un large panel d’études de cartographie des services écosystémiques
Hal-Diderot arrow_drop_down Mémoires en Sciences de l'Information et de la CommunicationDoctoral thesis . 2017All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______212::3f29e330e5fde3148e3f881329d8ee4c&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu0 citations 0 popularity Average influence Average impulse Average Powered by BIP!more_vert Hal-Diderot arrow_drop_down Mémoires en Sciences de l'Information et de la CommunicationDoctoral thesis . 2017All Research productsarrow_drop_down <script type="text/javascript"> <!-- document.write('<div id="oa_widget"></div>'); document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=od_______212::3f29e330e5fde3148e3f881329d8ee4c&type=result"></script>'); --> </script>
For further information contact us at helpdesk@openaire.eu