Digital Commons Network^™

Knowledge Representation And Reasoning With Deep Neural Networks, Arvind Ramanathan Neelakantan

Doctoral Dissertations

Knowledge representation and reasoning is one of the central challenges of artificial intelligence, and has important implications in many fields including natural language understanding and robotics. Representing knowledge with symbols, and reasoning via search and logic has been the dominant paradigm for many decades. In this work, we use deep neural networks to learn to both represent symbols and perform reasoning end-to-end from data. By learning powerful non-linear models, our approach generalizes to massive amounts of knowledge and works well with messy real-world data using minimal human effort. First, we show that recurrent neural networks with an attention mechanism achieve …

Temporal And Relational Models For Causality: Representation And Learning, Katerina Marazopoulou

Doctoral Dissertations

Discovering causal dependence is central to understanding the behavior of complex systems and to selecting actions that will achieve particular outcomes. The majority of work in this area has focused on propositional domains, where data instances are assumed to be independent and identically distributed (i.i.d.). However, many real-world domains are inherently relational, i.e., they consist of multiple types of entities that interact with each other, and temporal, i.e., they change over time. This thesis focuses on causal modeling for these more complex relational and temporal domains. This thesis provides an in-depth investigation of the properties of relational models and is …

Deep-Learned Generative Representations Of 3d Shape Families, Haibin Huang

Doctoral Dissertations

Digital representations of 3D shapes are becoming increasingly useful in several emerging applications, such as 3D printing, virtual reality and augmented reality. However, traditional modeling softwares require users to have extensive modeling experience, artistic skills and training to handle their complex interfaces and perform the necessary low-level geometric manipulation commands. Thus, there is an emerging need for computer algorithms that help novice and casual users to quickly and easily generate 3D content. In this work, I will present deep learning algorithms that are capable of automatically inferring parametric representations of shape families, which can be used to generate new 3D …

Database Usability Enhancement In Data Exploration, Yue Wang

Doctoral Dissertations

Database usability has become an important research topic over the last decade. In the early days, database management systems were maintained by sophisticated users like database administrators. Today, due to the availability of data and computing resources, more non-expert users are involved in database computation. From their point of view, database systems lack ease of use. So researchers believe that usability is as important as the performance and functionality of databases and therefore developed many techniques such as natural language interface to enhance the ease of use of databases. In this thesis, we find some deeper technical issues in database …

Adaft: A Resource-Efficient Framework For Adaptive Fault-Tolerance In Cyber-Physical Systems, Ye Xu

Doctoral Dissertations

Cyber-physical systems frequently have to use massive redundancy to meet application requirements for high reliability. While such redundancy is required, it can be activated adaptively, based on the current state of the controlled plant. Most of the time the physical plant is in a state that allows for a lower level of fault-tolerance. Avoiding the continuous deployment of massive fault-tolerance will greatly reduce the workload of CPSs. In this dissertation, we demonstrate a software simulation framework (AdaFT) that can automatically generate the sub-spaces within which our adaptive fault-tolerance can be applied. We also show the theoretical benefits of AdaFT, and …

Controversy Analysis And Detection, Shiri Dori-Hacohen

Doctoral Dissertations

Seeking information on a controversial topic is often a complex task. Alerting users about controversial search results can encourage critical literacy, promote healthy civic discourse and counteract the "filter bubble" effect, and therefore would be a useful feature in a search engine or browser extension. Additionally, presenting information to the user about the different stances or sides of the debate can help her navigate the landscape of search results beyond a simple "list of 10 links". This thesis has made strides in the emerging niche of controversy detection and analysis. The body of work in this thesis revolves around two …

The Complexity Of Resilience, Cibele Matos Freire

Doctoral Dissertations

One focus area in data management research is to understand how changes in the data can affect the output of a view or standing query. Example applications are explaining query results and propagating updates through views. In this thesis we study the complexity of the Resilience problem, which is the problem of finding the minimum number of tuples that need to be deleted from the database in order to change the result of a query. We will see that resilience is closely related to the well-studied problems of deletion propagation and causal responsibility, and that analyzing its complexity offers important …

Style-Driven Shape Analysis And Synthesis, Zhaoliang Lun

Doctoral Dissertations

In this dissertation I will investigate algorithms that analyze stylistic properties of 3D shapes and automatically synthesize shapes given style specifications. I will start by introducing a structure-transcending method for style similarity evaluation between 3D shapes. Inspired by observations about style similarity in art history literature, we propose an algorithmically computed style similarity measure which identifies style related elements on the analyzed models and collates element-level geometric similarity measurements into an object-level style measure consistent with human perception. To achieve this consistency we employ crowdsourcing to learn the relative perceptual importance of a range of elementary shape distances and other …

Graph Construction For Manifold Discovery, Cj Carey

Doctoral Dissertations

Manifold learning is a class of machine learning methods that exploits the observation that high-dimensional data tend to lie on a smooth lower-dimensional manifold. Manifold discovery is the essential first component of manifold learning methods, in which the manifold structure is inferred from available data. This task is typically posed as a graph construction problem: selecting a set of vertices and edges that most closely approximates the true underlying manifold. The quality of this learned graph is critical to the overall accuracy of the manifold learning method. Thus, it is essential to develop accurate, efficient, and reliable algorithms for constructing …

Spreadsheet Tools For Data Analysts, Daniel W. Barowy

Doctoral Dissertations

Spreadsheets are a natural fit for data analysis, combining a simple data storage and presentation layer with a programming language and basic debugging tools. Because spreadsheets are accessible and flexible, they are used by both novices and experts. Consequently, spreadsheets are hugely popular, with more than 750 million copies of Microsoft Excel installed worldwide. This popularity means that spreadsheets are the most popular programming language on the planet and the de facto tool for data analysis. Nevertheless, spreadsheets do not address a number of important tasks in a typical analyst's pipeline, and their design frequently complicates them. This thesis describes …

Deep Energy-Based Models For Structured Prediction, David Belanger

Doctoral Dissertations

We introduce structured prediction energy networks (SPENs), a flexible frame- work for structured prediction. A deep architecture is used to define an energy func- tion over candidate outputs and predictions are produced by gradient-based energy minimization. This deep energy captures dependencies between labels that would lead to intractable graphical models, and allows us to automatically discover discrim- inative features of the structured output. Furthermore, practitioners can explore a wide variety of energy function architectures without having to hand-design predic- tion and learning methods for each model. This is because all of our prediction and learning methods interact with the energy …

Towards Osgeo Best Practices For Scientific Software Citation: Integration Options For Persistent Identifiers Fn Osgeo Project Repositories, Peter Löwe, Markus Neteler, Jan Goebel, Marco Tullney

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

As a contribution to the currently ongoing larger effort to establish Open Science as best practices in academia, this article focuses on the Open Source and Open Access tiers of the Open Science triad and community software projects. The current situation of research software development and the need to recognize it as a significant contribution to science is introduced in relation to Open Science. The adoption of the Open Science paradigms occurs at different speeds and on different levels within the various fields of science and crosscutting software communities. This is paralleled by the emerging of an underlying futuresafe technical …

Using Osgeo Solutions For Local Development Systems Implementation. The Experience For The Northern Region Of Costa Rica, López-Villegas Oscar, Víquez-Acuña Oscar, Víquez-Acuña Leonardo

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

Although some general definitions classify Spatial Data Infrastructures (SDI) as technological standards, institutional and even political agreements, which allow the discovery and use of geospatial information by users for different purposes [Kuhn 2005], computationally this platforms are valuable data repositories that should reach people efficiently and effectively for analysis and decision making on issues of collective interest. Costa Rica has several SDIs experiences at national level (SNIT - http://www.snitcr.go.cr), regional level (IDEHN - http://www.idehn.tec.ac.cr) or local/cantonal level (IDESCA - http://idesca.cr). Those infrastructures can facilitate access between geospatial information managers and their consumers through the implementation of particular software applications. The …

Kadaster Data Platform - Overview Archicture, Erwin Folmer, Wouter Beek

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

The Dutch Cadastre is publishing its geospatial data assets as Linked Open Data through the Kadaster Data Platform (KDP). The KDP supports the following three Linked Data browsing paradigms: (1) graph navigation, (2) hierarchical browsing, and (3) faceted browsing. Graph navigation uses the graph-shape of the RDF datamodel, to display concepts and instances as nodes, and properties between them as edges between those nodes. Graph navigation works well for explorative browsing. For graph navigation the KDP uses LODView (http://lodview.it), an existing OSS. Hierarchical browsing uses the tree structure of the concept hierarchy in order to display the various classes and …

Evaluation Of The Micro-Tasking Method For Openstreetmap Imports, Atle Frenvik Sveen, Anne Sofie Strand Erichsen

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

Open Geospatial Data, capable of enriching OpenStreetMap, is being released by governments around the world at an increasing rate. The OSM import methods have been refined since the massive TIGER-import, moving towards assisted methods such as the with micro-tasking method used by the LA and NY buildings imports. While these imports serve as great case studies of imports, they do not deal with complex datasets, or updates to the data, neither do they deal with partitioning of tasks. We examine how the Norwegian FKB-dataset can be imported to OSM using micro-tasking, and perform a user-test to determine the best partition …

The Utility Of Beautiful Visualizations, Laura Tateosian, Reza Amindarbari, Christopher Healey, Pavel Kosik, James Enns

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

Geovisualizations provide a means to inspect large complex multivariate datasets for information that would not otherwise be available with a tabular view or summary statistics alone. Aesthetically appealing visualizations can elicit prolonged exploration and encourage discovery. Creating data geovisualizations that are effective and beautiful is an important yet difficult challenge. Here we present a tool for rendering geovisualizations of continuous spatial data using the painterly techniques of impressionist-era artists. The techniques, which have been tested in controlled studies, vary the visual properties (e.g., hue, size, and tilt) of brush strokes to represent multiple data attributes simultaneously in each location. To …

Tracking 19th Century Late Blight From Archival Documents Using Text Analytics And Geoparsing, Laura Tateosian, Rachael Guenter, Yi-Peng Yang, Jean Ristaino

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

In 1845, Ireland's potato crop was struck by a devastating potato disease that killed Ireland’s crop caused devastation for seven years and led to mass starvation and emigration from the country. The cause of the potato destruction was a fungus-like plant pathogen. There are several theories about the origin of the disease and the source of the 19th century outbreaks. We use historical documents contemporary to that time to investigate spatial information that might inform these mysteries. We present methodologies for automatically extracting information from these voluminous data sources. We identify and map geographic locations that are proximate in the …

Towards A Web-Enabled Geo-Sample Web: An Open Source Resource Registration And Management System For Connecting Geo-Samples To The Web, Anusuriya Devaraju, Jens Klump, Victor Tey, Simon Cox, Ryan Fraser

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

Within the earth sciences the curation and sharing of geo-samples is crucial to supporting reproducible research, in addition to extending the use of the samples in new research, and saving costs by avoiding sample loss and duplicating sampling activities. In the Commonwealth Scientific and Industrial Research Organisation (CSIRO), researchers gather various geo-samples as part of their field studies and collaborative projects. The diversity of the samples and their unsystematic management led ambiguous sample numbers, incomplete sample descriptions, and difficulties in finding the samples and their related data. These problems are also found in universities, research institutes and government agencies, which …

The Billion Object Platform (Bop): A System To Lower Barriers To Support Big, Streaming, Spatio-Temporal Data Sources, Devika Kakkar, Ben Lewis, David Smiley, Ariel Nunez

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

With funding from the Sloan Foundation and Harvard Dataverse, the Harvard Center for Geographic Analysis (CGA) has developed a big spatio-temporal data visualization platform called the Billion Object Platform or "BOP". The goal of the project is to lower barriers for scholars who wish to access large, streaming, spatio-temporal datasets. Since once archived, streaming data gets big fast, and since most GIS systems don't support interactive visualization of millions of objects, a new platform was needed. The BOP is loaded with the latest billion geo-tweets and is fed a real-time stream of about 1 million tweets per day. The CGA …

Processing Conservation Indicators With Open Source Tools: Lessons Learned From The Digital Observatory For Protected Areas, Lucy Bastin, Andrea Mandrici, Luca Battistella, Grégoire Dubois

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

The European Commission has a commitment to open data and the support of open source software and standards. We present lessons learnt while populating and supporting the web and map services that underly the Joint Research Centre's Digital Observatory for Protected Areas. Challenges include: large datasets with highly complex geometries; topological inconsistencies, compounded by reprojection for equal-area calculations; multiple different representations of the same geographical entities, for example coastlines; licensing requirement to continuously update indicators to respond to monthly changes in the authoritative data. In order to compute and publish an array of indicators, we used a range of open …

Optimizing Spatiotemporal Analysis Using Multidimensional Indexing With Geowave, Richard Fecher, Michael A. Whitby

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

The open source software GeoWave bridges the gap between geographic information systems and distributed computing. This is done by preserving locality of multidimensional data when indexing it into a single-dimensional key-value store, using space filling curves. This means that like values in each dimension are stored physically close together in the datastore. We demonstrate the efficiencies and benefits of the GeoWave indexing algorithm to store and query billions of spatiotemporal data points. We show how this indexing strategy can be used to reduce query and processing times by multiple orders of magnitude using publicly available taxi trip data published by …

Development Of An Extension Of Geoserver For Handling 3d Spatial Data, Hyung-Gyu Ryoo, Soojin Kim, Joon-Seok Kim, Ki-Joune Li

Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings

Recently, several open source software tools such as CesiumJS and iTowns have been developed for dealing with 3-dimensional spatial data. These tools mainly focus on visualization of 3D spatial data based on WebGL. An open-sourced server capable of storing, sharing and querying 3D spatial data has not yet been developed. GeoServer, one of the representative open source spatial data servers, provides many powerful features. In particular, it supports connecting to and publishing spatial data from a variety of data sources. GeoServer also supports Web Feature Service (WFS), which is a standard protocol established by the Open Geospatial Consortium to request …

High-Performance Complex Event Processing For Decision Analytics, Haopeng Zhang

Doctoral Dissertations

Complex Event Processing (CEP) systems are becoming increasingly popular in do- mains for decision analytics such as financial services, transportation, cluster monitoring, supply chain management, business process management, and health care. These systems collect or create high volumes event streams, and often require such event streams to be processed in real-time. To this end, CEP queries are applied for filtering, correlation, ag- gregation, and transformation, to derive high-level, actionable information. Tasks for CEP systems fall into two categories: passive monitoring and proactive monitoring. For passive monitoring, users know their exact needs and express them in CEP queries, then CEP engines …

Belief-Space Planning For Resourceful Manipulation And Mobility, Dirk Ruiken

Doctoral Dissertations

Robots are increasingly expected to work in partially observable and unstructured environments. They need to select actions that exploit perceptual and motor resourcefulness to manage uncertainty based on the demands of the task and environment. The research in this dissertation makes two primary contributions. First, it develops a new concept in resourceful robot platforms called the UMass uBot and introduces the sixth and seventh in the uBot series. uBot-6 introduces multiple postural configurations that enable different modes of mobility and manipulation to meet the needs of a wide variety of tasks and environmental constraints. uBot-7 extends this with the use …

Automatic Derivation Of Requirements For Components Used In Human-Intensive Systems, Heather Conboy

Doctoral Dissertations

Human-intensive systems (HISs), where humans must coordinate with each other along with software and/or hardware components to achieve system missions, are increasingly prevalent in safety-critical domains (e.g., healthcare). Such systems are often complex, involving aspects such as concurrency and exceptional situations. For these systems, it is often difficult but important to determine requirements for the individual components that are necessary to ensure the system requirements are satisfied. In this thesis, we investigated an approach that employs interface synthesis methods developed for software systems to automatically derive such requirements for components used in HISs. In previous work, we investigated a requirement …

Problems In Graph-Structured Modeling And Learning, James Atwood

Doctoral Dissertations

This thesis investigates three problems in graph-structured modeling and learning. We first present a method for efficiently generating large instances from nonlinear preferential attachment models of network structure. This is followed by a description of diffusion-convolutional neural networks, a new model for graph-structured data which is able to outperform probabilistic relational models and kernel-on-graph methods at node classification tasks. We conclude with an optimal privacy-protection method for users of online services that remains effective when users have poor knowledge of an adversary's behavior.

Method For Enabling Causal Inference In Relational Domains, David Arbour

Doctoral Dissertations

The analysis of data from complex systems is quickly becoming a fundamental aspect of modern business, government, and science. The field of causal learning is concerned with developing a set of statistical methods that allow practitioners make inferences about unseen interventions. This field has seen significant advances in recent years. However, the vast majority of this work assumes that data instances are independent, whereas many systems are best described in terms of interconnected instances, i.e. relational systems. This discrepancy prevents causal inference techniques from being reliably applied in many real-world settings.
In this thesis, I will present three contributions to …

Explorations Into Machine Learning Techniques For Precipitation Nowcasting, Aditya Nagarajan

Masters Theses

Recent advances in cloud-based big-data technologies now makes data driven solutions feasible for increasing numbers of scientific computing applications. One such data driven solution approach is machine learning where patterns in large data sets are brought to the surface by finding complex mathematical relationships within the data. Nowcasting or short-term prediction of rainfall in a given region is an important problem in meteorology. In this thesis we explore the nowcasting problem through a data driven approach by formulating it as a machine learning problem.

State-of-the-art nowcasting systems today are based on numerical models which describe the physical processes leading to …

Achieving Perfect Location Privacy In Wireless Devices Using Anonymization, Zarrin Montazeri

Masters Theses

The popularity of mobile devices and location-based services (LBS) have created great concerns regarding the location privacy of the users of such devices and services. Anonymization is a common technique that is often being used to protect the location privacy of LBS users. This technique assigns a random pseudonym to each user and these pseudonyms can change over time. Here, we provide a general information theoretic definition for perfect location privacy and prove that perfect location privacy is achievable for mobile devices when using the anonymization technique appropriately. First, we assume that the user’s current location is independent from her …

On Leveraging Multi-Path Transport In Mobile Networks, Yeon-Sup Lim

Doctoral Dissertations

Multi-Path TCP (MPTCP) is a new transport protocol that enables mobile devices to simultaneously use several physical paths through multiple network interfaces. MPTCP is particularly useful for mobile devices, which usually have multiple wireless interfaces such as IEEE 802.11 (WiFi), cellular (3G/LTE), and Bluetooth. However, applying MPTCP to mobile devices introduces new concerns since they operate in harsh environments with resource constraints due to intermittent path availability and limited power supply. The goal of this thesis is to resolve these problems so as to be able to practically deploy MPTCP in mobile devices. The first part of the thesis develops …