Matching Items (400)
Filtering by
- All Subjects: Computer Science
- Genre: Doctoral Dissertation
- Creators: Arizona State University
Description
Many learning models have been proposed for various tasks in visual computing. Popular examples include hidden Markov models and support vector machines. Recently, sparse-representation-based learning methods have attracted a lot of attention in the computer vision field, largely because of their impressive performance in many applications. In the literature, many of such sparse learning methods focus on designing or application of some learning techniques for certain feature space without much explicit consideration on possible interaction between the underlying semantics of the visual data and the employed learning technique. Rich semantic information in most visual data, if properly incorporated into algorithm design, should help achieving improved performance while delivering intuitive interpretation of the algorithmic outcomes. My study addresses the problem of how to explicitly consider the semantic information of the visual data in the sparse learning algorithms. In this work, we identify four problems which are of great importance and broad interest to the community. Specifically, a novel approach is proposed to incorporate label information to learn a dictionary which is not only reconstructive but also discriminative; considering the formation process of face images, a novel image decomposition approach for an ensemble of correlated images is proposed, where a subspace is built from the decomposition and applied to face recognition; based on the observation that, the foreground (or salient) objects are sparse in input domain and the background is sparse in frequency domain, a novel and efficient spatio-temporal saliency detection algorithm is proposed to identify the salient regions in video; and a novel hidden Markov model learning approach is proposed by utilizing a sparse set of pairwise comparisons among the data, which is easier to obtain and more meaningful, consistent than tradition labels, in many scenarios, e.g., evaluating motion skills in surgical simulations. In those four problems, different types of semantic information are modeled and incorporated in designing sparse learning algorithms for the corresponding visual computing tasks. Several real world applications are selected to demonstrate the effectiveness of the proposed methods, including, face recognition, spatio-temporal saliency detection, abnormality detection, spatio-temporal interest point detection, motion analysis and emotion recognition. In those applications, data of different modalities are involved, ranging from audio signal, image to video. Experiments on large scale real world data with comparisons to state-of-art methods confirm the proposed approaches deliver salient advantages, showing adding those semantic information dramatically improve the performances of the general sparse learning methods.
ContributorsZhang, Qiang (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Wang, Yalin (Committee member) / Ye, Jieping (Committee member) / Arizona State University (Publisher)
Created2014
Description
Audio signals, such as speech and ambient sounds convey rich information pertaining to a user’s activity, mood or intent. Enabling machines to understand this contextual information is necessary to bridge the gap in human-machine interaction. This is challenging due to its subjective nature, hence, requiring sophisticated techniques. This dissertation presents a set of computational methods, that generalize well across different conditions, for speech-based applications involving emotion recognition and keyword detection, and ambient sounds-based applications such as lifelogging.
The expression and perception of emotions varies across speakers and cultures, thus, determining features and classification methods that generalize well to different conditions is strongly desired. A latent topic models-based method is proposed to learn supra-segmental features from low-level acoustic descriptors. The derived features outperform state-of-the-art approaches over multiple databases. Cross-corpus studies are conducted to determine the ability of these features to generalize well across different databases. The proposed method is also applied to derive features from facial expressions; a multi-modal fusion overcomes the deficiencies of a speech only approach and further improves the recognition performance.
Besides affecting the acoustic properties of speech, emotions have a strong influence over speech articulation kinematics. A learning approach, which constrains a classifier trained over acoustic descriptors, to also model articulatory data is proposed here. This method requires articulatory information only during the training stage, thus overcoming the challenges inherent to large-scale data collection, while simultaneously exploiting the correlations between articulation kinematics and acoustic descriptors to improve the accuracy of emotion recognition systems.
Identifying context from ambient sounds in a lifelogging scenario requires feature extraction, segmentation and annotation techniques capable of efficiently handling long duration audio recordings; a complete framework for such applications is presented. The performance is evaluated on real world data and accompanied by a prototypical Android-based user interface.
The proposed methods are also assessed in terms of computation and implementation complexity. Software and field programmable gate array based implementations are considered for emotion recognition, while virtual platforms are used to model the complexities of lifelogging. The derived metrics are used to determine the feasibility of these methods for applications requiring real-time capabilities and low power consumption.
The expression and perception of emotions varies across speakers and cultures, thus, determining features and classification methods that generalize well to different conditions is strongly desired. A latent topic models-based method is proposed to learn supra-segmental features from low-level acoustic descriptors. The derived features outperform state-of-the-art approaches over multiple databases. Cross-corpus studies are conducted to determine the ability of these features to generalize well across different databases. The proposed method is also applied to derive features from facial expressions; a multi-modal fusion overcomes the deficiencies of a speech only approach and further improves the recognition performance.
Besides affecting the acoustic properties of speech, emotions have a strong influence over speech articulation kinematics. A learning approach, which constrains a classifier trained over acoustic descriptors, to also model articulatory data is proposed here. This method requires articulatory information only during the training stage, thus overcoming the challenges inherent to large-scale data collection, while simultaneously exploiting the correlations between articulation kinematics and acoustic descriptors to improve the accuracy of emotion recognition systems.
Identifying context from ambient sounds in a lifelogging scenario requires feature extraction, segmentation and annotation techniques capable of efficiently handling long duration audio recordings; a complete framework for such applications is presented. The performance is evaluated on real world data and accompanied by a prototypical Android-based user interface.
The proposed methods are also assessed in terms of computation and implementation complexity. Software and field programmable gate array based implementations are considered for emotion recognition, while virtual platforms are used to model the complexities of lifelogging. The derived metrics are used to determine the feasibility of these methods for applications requiring real-time capabilities and low power consumption.
ContributorsShah, Mohit (Author) / Spanias, Andreas (Thesis advisor) / Chakrabarti, Chaitali (Thesis advisor) / Berisha, Visar (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2015
Description
Crises or large-scale emergencies such as earthquakes and hurricanes cause massive damage to lives and property. Crisis response is an essential task to mitigate the impact of a crisis. An effective response to a crisis necessitates information gathering and analysis. Traditionally, this process has been restricted to the information collected by first responders on the ground in the affected region or by official agencies such as local governments involved in the response. However, the ubiquity of mobile devices has empowered people to publish information during a crisis through social media, such as the damage reports from a hurricane. Social media has thus emerged as an important channel of information which can be leveraged to improve crisis response. Twitter is a popular medium which has been employed in recent crises. However, it presents new challenges: the data is noisy and uncurated, and it has high volume and high velocity. In this work, I study four key problems in the use of social media for crisis response: effective monitoring and analysis of high volume crisis tweets, detecting crisis events automatically in streaming data, identifying users who can be followed to effectively monitor crisis, and finally understanding user behavior during crisis to detect tweets inside crisis regions. To address these problems I propose two systems which assist disaster responders or analysts to collaboratively collect tweets related to crisis and analyze it using visual analytics to identify interesting regions, topics, and users involved in disaster response. I present a novel approach to detecting crisis events automatically in noisy, high volume Twitter streams. I also investigate and introduce novel methods to tackle information overload through the identification of information leaders in information diffusion who can be followed for efficient crisis monitoring and identification of messages originating from crisis regions using user behavior analysis.
ContributorsKumar, Shamanth (Author) / Liu, Huan (Thesis advisor) / Davulcu, Hasan (Committee member) / Maciejewski, Ross (Committee member) / Agarwal, Nitin (Committee member) / Arizona State University (Publisher)
Created2015
Description
Social networking services have emerged as an important platform for large-scale information sharing and communication. With the growing popularity of social media, spamming has become rampant in the platforms. Complex network interactions and evolving content present great challenges for social spammer detection. Different from some existing well-studied platforms, distinct characteristics of newly emerged social media data present new challenges for social spammer detection. First, texts in social media are short and potentially linked with each other via user connections. Second, it is observed that abundant contextual information may play an important role in distinguishing social spammers and normal users. Third, not only the content information but also the social connections in social media evolve very fast. Fourth, it is easy to amass vast quantities of unlabeled data in social media, but would be costly to obtain labels, which are essential for many supervised algorithms. To tackle those challenges raise in social media data, I focused on developing effective and efficient machine learning algorithms for social spammer detection.
I provide a novel and systematic study of social spammer detection in the dissertation. By analyzing the properties of social network and content information, I propose a unified framework for social spammer detection by collectively using the two types of information in social media. Motivated by psychological findings in physical world, I investigate whether sentiment analysis can help spammer detection in online social media. In particular, I conduct an exploratory study to analyze the sentiment differences between spammers and normal users; and present a novel method to incorporate sentiment information into social spammer detection framework. Given the rapidly evolving nature, I propose a novel framework to efficiently reflect the effect of newly emerging social spammers. To tackle the problem of lack of labeling data in social media, I study how to incorporate network information into text content modeling, and design strategies to select the most representative and informative instances from social media for labeling. Motivated by publicly available label information from other media platforms, I propose to make use of knowledge learned from cross-media to help spammer detection on social media.
I provide a novel and systematic study of social spammer detection in the dissertation. By analyzing the properties of social network and content information, I propose a unified framework for social spammer detection by collectively using the two types of information in social media. Motivated by psychological findings in physical world, I investigate whether sentiment analysis can help spammer detection in online social media. In particular, I conduct an exploratory study to analyze the sentiment differences between spammers and normal users; and present a novel method to incorporate sentiment information into social spammer detection framework. Given the rapidly evolving nature, I propose a novel framework to efficiently reflect the effect of newly emerging social spammers. To tackle the problem of lack of labeling data in social media, I study how to incorporate network information into text content modeling, and design strategies to select the most representative and informative instances from social media for labeling. Motivated by publicly available label information from other media platforms, I propose to make use of knowledge learned from cross-media to help spammer detection on social media.
ContributorsHu, Xia, Ph.D (Author) / Liu, Huan (Thesis advisor) / Kambhampati, Subbarao (Committee member) / Ye, Jieping (Committee member) / Faloutsos, Christos (Committee member) / Arizona State University (Publisher)
Created2015
Description
A myriad of social media services are emerging in recent years that allow people to communicate and express themselves conveniently and easily. The pervasive use of social media generates massive data at an unprecedented rate. It becomes increasingly difficult for online users to find relevant information or, in other words, exacerbates the information overload problem. Meanwhile, users in social media can be both passive content consumers and active content producers, causing the quality of user-generated content can vary dramatically from excellence to abuse or spam, which results in a problem of information credibility. Trust, providing evidence about with whom users can trust to share information and from whom users can accept information without additional verification, plays a crucial role in helping online users collect relevant and reliable information. It has been proven to be an effective way to mitigate information overload and credibility problems and has attracted increasing attention.
As the conceptual counterpart of trust, distrust could be as important as trust and its value has been widely recognized by social sciences in the physical world. However, little attention is paid on distrust in social media. Social media differs from the physical world - (1) its data is passively observed, large-scale, incomplete, noisy and embedded with rich heterogeneous sources; and (2) distrust is generally unavailable in social media. These unique properties of social media present novel challenges for computing distrust in social media: (1) passively observed social media data does not provide necessary information social scientists use to understand distrust, how can I understand distrust in social media? (2) distrust is usually invisible in social media, how can I make invisible distrust visible by leveraging unique properties of social media data? and (3) little is known about distrust and its role in social media applications, how can distrust help make difference in social media applications?
The chief objective of this dissertation is to figure out solutions to these challenges via innovative research and novel methods. In particular, computational tasks are designed to {\it understand distrust}, a innovative task, i.e., {\it predicting distrust} is proposed with novel frameworks to make invisible distrust visible, and principled approaches are develop to {\it apply distrust} in social media applications. Since distrust is a special type of negative links, I demonstrate the generalization of properties and algorithms of distrust to negative links, i.e., {\it generalizing findings of distrust}, which greatly expands the boundaries of research of distrust and largely broadens its applications in social media.
As the conceptual counterpart of trust, distrust could be as important as trust and its value has been widely recognized by social sciences in the physical world. However, little attention is paid on distrust in social media. Social media differs from the physical world - (1) its data is passively observed, large-scale, incomplete, noisy and embedded with rich heterogeneous sources; and (2) distrust is generally unavailable in social media. These unique properties of social media present novel challenges for computing distrust in social media: (1) passively observed social media data does not provide necessary information social scientists use to understand distrust, how can I understand distrust in social media? (2) distrust is usually invisible in social media, how can I make invisible distrust visible by leveraging unique properties of social media data? and (3) little is known about distrust and its role in social media applications, how can distrust help make difference in social media applications?
The chief objective of this dissertation is to figure out solutions to these challenges via innovative research and novel methods. In particular, computational tasks are designed to {\it understand distrust}, a innovative task, i.e., {\it predicting distrust} is proposed with novel frameworks to make invisible distrust visible, and principled approaches are develop to {\it apply distrust} in social media applications. Since distrust is a special type of negative links, I demonstrate the generalization of properties and algorithms of distrust to negative links, i.e., {\it generalizing findings of distrust}, which greatly expands the boundaries of research of distrust and largely broadens its applications in social media.
ContributorsTang, Jiliang (Author) / Liu, Huan (Thesis advisor) / Xue, Guoliang (Committee member) / Ye, Jieping (Committee member) / Aggarwal, Charu (Committee member) / Arizona State University (Publisher)
Created2015
Description
While discrete emotions like joy, anger, disgust etc. are quite popular, continuous
emotion dimensions like arousal and valence are gaining popularity within the research
community due to an increase in the availability of datasets annotated with these
emotions. Unlike the discrete emotions, continuous emotions allow modeling of subtle
and complex affect dimensions but are difficult to predict.
Dimension reduction techniques form the core of emotion recognition systems and
help create a new feature space that is more helpful in predicting emotions. But these
techniques do not necessarily guarantee a better predictive capability as most of them
are unsupervised, especially in regression learning. In emotion recognition literature,
supervised dimension reduction techniques have not been explored much and in this
work a solution is provided through probabilistic topic models. Topic models provide
a strong probabilistic framework to embed new learning paradigms and modalities.
In this thesis, the graphical structure of Latent Dirichlet Allocation has been explored
and new models tuned to emotion recognition and change detection have been built.
In this work, it has been shown that the double mixture structure of topic models
helps 1) to visualize feature patterns, and 2) to project features onto a topic simplex
that is more predictive of human emotions, when compared to popular techniques
like PCA and KernelPCA. Traditionally, topic models have been used on quantized
features but in this work, a continuous topic model called the Dirichlet Gaussian
Mixture model has been proposed. Evaluation of DGMM has shown that while modeling
videos, performance of LDA models can be replicated even without quantizing
the features. Until now, topic models have not been explored in a supervised context
of video analysis and thus a Regularized supervised topic model (RSLDA) that
models video and audio features is introduced. RSLDA learning algorithm performs
both dimension reduction and regularized linear regression simultaneously, and has outperformed supervised dimension reduction techniques like SPCA and Correlation
based feature selection algorithms. In a first of its kind, two new topic models, Adaptive
temporal topic model (ATTM) and SLDA for change detection (SLDACD) have
been developed for predicting concept drift in time series data. These models do not
assume independence of consecutive frames and outperform traditional topic models
in detecting local and global changes respectively.
emotion dimensions like arousal and valence are gaining popularity within the research
community due to an increase in the availability of datasets annotated with these
emotions. Unlike the discrete emotions, continuous emotions allow modeling of subtle
and complex affect dimensions but are difficult to predict.
Dimension reduction techniques form the core of emotion recognition systems and
help create a new feature space that is more helpful in predicting emotions. But these
techniques do not necessarily guarantee a better predictive capability as most of them
are unsupervised, especially in regression learning. In emotion recognition literature,
supervised dimension reduction techniques have not been explored much and in this
work a solution is provided through probabilistic topic models. Topic models provide
a strong probabilistic framework to embed new learning paradigms and modalities.
In this thesis, the graphical structure of Latent Dirichlet Allocation has been explored
and new models tuned to emotion recognition and change detection have been built.
In this work, it has been shown that the double mixture structure of topic models
helps 1) to visualize feature patterns, and 2) to project features onto a topic simplex
that is more predictive of human emotions, when compared to popular techniques
like PCA and KernelPCA. Traditionally, topic models have been used on quantized
features but in this work, a continuous topic model called the Dirichlet Gaussian
Mixture model has been proposed. Evaluation of DGMM has shown that while modeling
videos, performance of LDA models can be replicated even without quantizing
the features. Until now, topic models have not been explored in a supervised context
of video analysis and thus a Regularized supervised topic model (RSLDA) that
models video and audio features is introduced. RSLDA learning algorithm performs
both dimension reduction and regularized linear regression simultaneously, and has outperformed supervised dimension reduction techniques like SPCA and Correlation
based feature selection algorithms. In a first of its kind, two new topic models, Adaptive
temporal topic model (ATTM) and SLDA for change detection (SLDACD) have
been developed for predicting concept drift in time series data. These models do not
assume independence of consecutive frames and outperform traditional topic models
in detecting local and global changes respectively.
ContributorsLade, Prasanth (Author) / Panchanathan, Sethuraman (Thesis advisor) / Davulcu, Hasan (Committee member) / Li, Baoxin (Committee member) / Balasubramanian, Vineeth N (Committee member) / Arizona State University (Publisher)
Created2015
Description
Resource allocation is one of the most challenging issues policy decision makers must address. The objective of this thesis is to explore the resource allocation from an economical perspective, i.e., how to purchase resources in order to satisfy customers' requests. In this thesis, we attend to answer the question: when and how to buy resources to fulfill customers' demands with minimum costs?
The first topic studied in this thesis is resource allocation in cloud networks. Cloud computing heralded an era where resources (such as computation and storage) can be scaled up and down elastically and on demand. This flexibility is attractive for its cost effectiveness: the cloud resource price depends on the actual utilization over time. This thesis studies two critical problems in cloud networks, focusing on the economical aspects of the resource allocation in the cloud/virtual networks, and proposes six algorithms to address the resource allocation problems for different discount models. The first problem attends a scenario where the virtual network provider offers different contracts to the service provider. Four algorithms for resource contract migration are proposed under two pricing models: Pay-as-You-Come and Pay-as-You-Go. The second problem explores a scenario where a cloud provider offers k contracts each with a duration and a rate respectively and a customer buys these contracts in order to satisfy its resource demand. This work shows that this problem can be seen as a 2-dimensional generalization of the classic online parking permit problem, and present a k-competitive online algorithm and an optimal online algorithm.
The second topic studied in this thesis is to explore how resource allocation and purchasing strategies work in our daily life. For example, is it worth buying a Yoga pass which costs USD 100 for ten entries, although it will expire at the end of this year? Decisions like these are part of our daily life, yet, not much is known today about good online strategies to buy discount vouchers with expiration dates. This work hence introduces a Discount Voucher Purchase Problem (DVPP). It aims to optimize the strategies for buying discount vouchers, i.e., coupons, vouchers, groupons which are valid only during a certain time period. The DVPP comes in three flavors: (1) Once Expire Lose Everything (OELE): Vouchers lose their entire value after expiration. (2) Once Expire Lose Discount (OELD): Vouchers lose their discount value after expiration. (3) Limited Purchasing Window (LPW): Vouchers have the property of OELE and can only be bought during a certain time window.
This work explores online algorithms with a provable competitive ratio against a clairvoyant offline algorithm, even in the worst case. In particular, this work makes the following contributions: we present a 4-competitive algorithm for OELE, an 8-competitive algorithm for OELD, and a lower bound for LPW. We also present an optimal offline algorithm for OELE and LPW, and show it is a 2-approximation solution for OELD.
The first topic studied in this thesis is resource allocation in cloud networks. Cloud computing heralded an era where resources (such as computation and storage) can be scaled up and down elastically and on demand. This flexibility is attractive for its cost effectiveness: the cloud resource price depends on the actual utilization over time. This thesis studies two critical problems in cloud networks, focusing on the economical aspects of the resource allocation in the cloud/virtual networks, and proposes six algorithms to address the resource allocation problems for different discount models. The first problem attends a scenario where the virtual network provider offers different contracts to the service provider. Four algorithms for resource contract migration are proposed under two pricing models: Pay-as-You-Come and Pay-as-You-Go. The second problem explores a scenario where a cloud provider offers k contracts each with a duration and a rate respectively and a customer buys these contracts in order to satisfy its resource demand. This work shows that this problem can be seen as a 2-dimensional generalization of the classic online parking permit problem, and present a k-competitive online algorithm and an optimal online algorithm.
The second topic studied in this thesis is to explore how resource allocation and purchasing strategies work in our daily life. For example, is it worth buying a Yoga pass which costs USD 100 for ten entries, although it will expire at the end of this year? Decisions like these are part of our daily life, yet, not much is known today about good online strategies to buy discount vouchers with expiration dates. This work hence introduces a Discount Voucher Purchase Problem (DVPP). It aims to optimize the strategies for buying discount vouchers, i.e., coupons, vouchers, groupons which are valid only during a certain time period. The DVPP comes in three flavors: (1) Once Expire Lose Everything (OELE): Vouchers lose their entire value after expiration. (2) Once Expire Lose Discount (OELD): Vouchers lose their discount value after expiration. (3) Limited Purchasing Window (LPW): Vouchers have the property of OELE and can only be bought during a certain time window.
This work explores online algorithms with a provable competitive ratio against a clairvoyant offline algorithm, even in the worst case. In particular, this work makes the following contributions: we present a 4-competitive algorithm for OELE, an 8-competitive algorithm for OELD, and a lower bound for LPW. We also present an optimal offline algorithm for OELE and LPW, and show it is a 2-approximation solution for OELD.
ContributorsHu, Xinhui (Author) / Richa, Andrea (Thesis advisor) / Schmid, Stefan (Committee member) / Sen, Arunabha (Committee member) / Xue, Guoliang (Committee member) / Arizona State University (Publisher)
Created2015
Description
US Senate is the venue of political debates where the federal bills are formed and voted. Senators show their support/opposition along the bills with their votes. This information makes it possible to extract the polarity of the senators. Similarly, blogosphere plays an increasingly important role as a forum for public debate. Authors display sentiment toward issues, organizations or people using a natural language.
In this research, given a mixed set of senators/blogs debating on a set of political issues from opposing camps, I use signed bipartite graphs for modeling debates, and I propose an algorithm for partitioning both the opinion holders (senators or blogs) and the issues (bills or topics) comprising the debate into binary opposing camps. Simultaneously, my algorithm scales the entities on a univariate scale. Using this scale, a researcher can identify moderate and extreme senators/blogs within each camp, and polarizing versus unifying issues. Through performance evaluations I show that my proposed algorithm provides an effective solution to the problem, and performs much better than existing baseline algorithms adapted to solve this new problem. In my experiments, I used both real data from political blogosphere and US Congress records, as well as synthetic data which were obtained by varying polarization and degree distribution of the vertices of the graph to show the robustness of my algorithm.
I also applied my algorithm on all the terms of the US Senate to the date for longitudinal analysis and developed a web based interactive user interface www.PartisanScale.com to visualize the analysis.
US politics is most often polarized with respect to the left/right alignment of the entities. However, certain issues do not reflect the polarization due to political parties, but observe a split correlating to the demographics of the senators, or simply receive consensus. I propose a hierarchical clustering algorithm that identifies groups of bills that share the same polarization characteristics. I developed a web based interactive user interface www.ControversyAnalysis.com to visualize the clusters while providing a synopsis through distribution charts, word clouds, and heat maps.
In this research, given a mixed set of senators/blogs debating on a set of political issues from opposing camps, I use signed bipartite graphs for modeling debates, and I propose an algorithm for partitioning both the opinion holders (senators or blogs) and the issues (bills or topics) comprising the debate into binary opposing camps. Simultaneously, my algorithm scales the entities on a univariate scale. Using this scale, a researcher can identify moderate and extreme senators/blogs within each camp, and polarizing versus unifying issues. Through performance evaluations I show that my proposed algorithm provides an effective solution to the problem, and performs much better than existing baseline algorithms adapted to solve this new problem. In my experiments, I used both real data from political blogosphere and US Congress records, as well as synthetic data which were obtained by varying polarization and degree distribution of the vertices of the graph to show the robustness of my algorithm.
I also applied my algorithm on all the terms of the US Senate to the date for longitudinal analysis and developed a web based interactive user interface www.PartisanScale.com to visualize the analysis.
US politics is most often polarized with respect to the left/right alignment of the entities. However, certain issues do not reflect the polarization due to political parties, but observe a split correlating to the demographics of the senators, or simply receive consensus. I propose a hierarchical clustering algorithm that identifies groups of bills that share the same polarization characteristics. I developed a web based interactive user interface www.ControversyAnalysis.com to visualize the clusters while providing a synopsis through distribution charts, word clouds, and heat maps.
ContributorsGokalp, Sedat (Author) / Davulcu, Hasan (Thesis advisor) / Sen, Arunabha (Committee member) / Liu, Huan (Committee member) / Woodward, Mark (Committee member) / Arizona State University (Publisher)
Created2015
Description
This thesis focuses on generating and exploring design variations for architectural and urban layouts. I propose to study this general problem in three selected contexts.
First, I introduce a framework to generate many variations of a facade design that look similar to a given facade layout. Starting from an input image, the facade is hierarchically segmented and labeled with a collection of manual and automatic tools. The user can then model constraints that should be maintained in any variation of the input facade design. Subsequently, facade variations are generated for different facade sizes, where multiple variations can be produced for a certain size.
Second, I propose a method for a user to understand and systematically explore good building layouts. Starting from a discrete set of good layouts, I analytically characterize the local shape space of good layouts around each initial layout, compactly encode these spaces, and link them to support transitions across the different local spaces. I represent such transitions in the form of a portal graph. The user can then use the portal graph, along with the family of local shape spaces, to globally and locally explore the space of good building layouts.
Finally, I propose an algorithm to computationally design street networks that balance competing requirements such as quick travel time and reduced through traffic in residential neighborhoods. The user simply provides high-level functional specifications for a target neighborhood, while my algorithm best satisfies the specification by solving for both connectivity and geometric layout of the network.
First, I introduce a framework to generate many variations of a facade design that look similar to a given facade layout. Starting from an input image, the facade is hierarchically segmented and labeled with a collection of manual and automatic tools. The user can then model constraints that should be maintained in any variation of the input facade design. Subsequently, facade variations are generated for different facade sizes, where multiple variations can be produced for a certain size.
Second, I propose a method for a user to understand and systematically explore good building layouts. Starting from a discrete set of good layouts, I analytically characterize the local shape space of good layouts around each initial layout, compactly encode these spaces, and link them to support transitions across the different local spaces. I represent such transitions in the form of a portal graph. The user can then use the portal graph, along with the family of local shape spaces, to globally and locally explore the space of good building layouts.
Finally, I propose an algorithm to computationally design street networks that balance competing requirements such as quick travel time and reduced through traffic in residential neighborhoods. The user simply provides high-level functional specifications for a target neighborhood, while my algorithm best satisfies the specification by solving for both connectivity and geometric layout of the network.
ContributorsBao, Fan (Author) / Wonka, Peter (Thesis advisor) / Maciejewski, Ross (Committee member) / Razdan, Anshuman (Committee member) / Farin, Gerald (Committee member) / Arizona State University (Publisher)
Created2014
Description
Stream processing has emerged as an important model of computation especially in the context of multimedia and communication sub-systems of embedded System-on-Chip (SoC) architectures. The dataflow nature of streaming applications allows them to be most naturally expressed as a set of kernels iteratively operating on continuous streams of data. The kernels are computationally intensive and are mainly characterized by real-time constraints that demand high throughput and data bandwidth with limited global data reuse. Conventional architectures fail to meet these demands due to their poorly matched execution models and the overheads associated with instruction and data movements.
This work presents StreamWorks, a multi-core embedded architecture for energy-efficient stream computing. The basic processing element in the StreamWorks architecture is the StreamEngine (SE) which is responsible for iteratively executing a stream kernel. SE introduces an instruction locking mechanism that exploits the iterative nature of the kernels and enables fine-grain instruction reuse. Each instruction in a SE is locked to a Reservation Station (RS) and revitalizes itself after execution; thus never retiring from the RS. The entire kernel is hosted in RS Banks (RSBs) close to functional units for energy-efficient instruction delivery. The dataflow semantics of stream kernels are captured by a context-aware dataflow execution mode that efficiently exploits the Instruction Level Parallelism (ILP) and Data-level parallelism (DLP) within stream kernels.
Multiple SEs are grouped together to form a StreamCluster (SC) that communicate via a local interconnect. A novel software FIFO virtualization technique with split-join functionality is proposed for efficient and scalable stream communication across SEs. The proposed communication mechanism exploits the Task-level parallelism (TLP) of the stream application. The performance and scalability of the communication mechanism is evaluated against the existing data movement schemes for scratchpad based multi-core architectures. Further, overlay schemes and architectural support are proposed that allow hosting any number of kernels on the StreamWorks architecture. The proposed oevrlay schemes for code management supports kernel(context) switching for the most common use cases and can be adapted for any multi-core architecture that use software managed local memories.
The performance and energy-efficiency of the StreamWorks architecture is evaluated for stream kernel and application benchmarks by implementing the architecture in 45nm TSMC and comparison with a low power RISC core and a contemporary accelerator.
This work presents StreamWorks, a multi-core embedded architecture for energy-efficient stream computing. The basic processing element in the StreamWorks architecture is the StreamEngine (SE) which is responsible for iteratively executing a stream kernel. SE introduces an instruction locking mechanism that exploits the iterative nature of the kernels and enables fine-grain instruction reuse. Each instruction in a SE is locked to a Reservation Station (RS) and revitalizes itself after execution; thus never retiring from the RS. The entire kernel is hosted in RS Banks (RSBs) close to functional units for energy-efficient instruction delivery. The dataflow semantics of stream kernels are captured by a context-aware dataflow execution mode that efficiently exploits the Instruction Level Parallelism (ILP) and Data-level parallelism (DLP) within stream kernels.
Multiple SEs are grouped together to form a StreamCluster (SC) that communicate via a local interconnect. A novel software FIFO virtualization technique with split-join functionality is proposed for efficient and scalable stream communication across SEs. The proposed communication mechanism exploits the Task-level parallelism (TLP) of the stream application. The performance and scalability of the communication mechanism is evaluated against the existing data movement schemes for scratchpad based multi-core architectures. Further, overlay schemes and architectural support are proposed that allow hosting any number of kernels on the StreamWorks architecture. The proposed oevrlay schemes for code management supports kernel(context) switching for the most common use cases and can be adapted for any multi-core architecture that use software managed local memories.
The performance and energy-efficiency of the StreamWorks architecture is evaluated for stream kernel and application benchmarks by implementing the architecture in 45nm TSMC and comparison with a low power RISC core and a contemporary accelerator.
ContributorsPanda, Amrit (Author) / Chatha, Karam S. (Thesis advisor) / Wu, Carole-Jean (Thesis advisor) / Chakrabarti, Chaitali (Committee member) / Shrivastava, Aviral (Committee member) / Arizona State University (Publisher)
Created2014