CS Theses & Dissertations 2014
For 2014 graduation dates (in alphabetical order by last name):
Distributed skip list in fine-grain message passing interface : implementation and analysis of a dictionary data structure that supports range queries
Alam, Sarwar
DOI : 10.14288/1.0166876
URI : http://hdl.handle.net/2429/46286
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Wagner
Interactive visualization for group decision-making
Bajracharya, Sanjana
DOI : 10.14288/1.0166963
URI : http://hdl.handle.net/2429/50262
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Poole
Implementation and evaluation of a classroom synchronous participation system
Beshai, Peter
DOI : 10.14288/1.0166030
URI : http://hdl.handle.net/2429/50165
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Booth
Structural comparison of source code between multiple programming languages
Biehn, Rolf Armin
DOI : 10.14288/1.0103428
URI : http://hdl.handle.net/2429/46547
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Wohlstadter
Eye-tracking as a source of information for automatically predicting user learning with MetaTutor, an intelligent tutoring system to support self-regulated learning
Bondareva, Daria
DOI : 10.14288/1.0166864
URI : http://hdl.handle.net/2429/46258
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Conati
Query-driven event search in news information network
Chen, Shanshan
DOI : 10.14288/1.0165990
URI : http://hdl.handle.net/2429/50065
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Lakshmanan
A search set model of path tracing in graphs
Dawson, Jessica Quinn
DOI : 10.14288/1.0052172
URI : http://hdl.handle.net/2429/45404
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisors : Dr. McGrenere, Dr. Munzner
Lifespace tracking and activity monitoring on mobile phones
Dewancker, Ian Richard
DOI : 10.14288/1.0166870
URI : http://hdl.handle.net/2429/46269
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Mitchell
Sensing and sorting ore using a relational influence diagram
Dirks, Matthew
DOI : 10.14288/1.0165936
URI : http://hdl.handle.net/2429/49998
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Poole
A streaming algorithms approach to approximating hit rate curves
Drudi, Zachary
DOI : 10.14288/1.0166992
URI : http://hdl.handle.net/2429/50486
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Harvey
Control of complex biomechanical systems
Fain, Mikhail
DOI : 10.14288/1.0052179
URI : http://hdl.handle.net/2429/45401
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Pai
Precision manipulations using a low-dimensional haptic interface
Humberston, Benjamin James
DOI : 10.14288/1.0165938
URI : http://hdl.handle.net/2429/49994
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Pai
Evaluating open relation extraction over conversational texts
Imani, Mahsa
DOI : 10.14288/1.0165856
URI : http://hdl.handle.net/2429/45978
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Carenini
Fault tolerance for distributed explicit-state model checking
Ishida, Valerie Lynn
DOI : 10.14288/1.0167008
URI : http://hdl.handle.net/2429/50746
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Greenstreet
Predicting affect in an Intelligent Tutoring System
Jaques, Natasha Mary
DOI : 10.14288/1.0135541
URI : http://hdl.handle.net/2429/50291
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Conati
Explanations for command recommendations : an experimental study
Jiresal, Rahul Sudhir
DOI : 10.14288/1.0165690
URI : http://hdl.handle.net/2429/45611
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. G. Murphy
Discourse analysis of asynchronous conversations
Joty, Shafiq Rayhan
DOI : 10.14288/1.0165726
URI : http://hdl.handle.net/2429/45674
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-05
Supervisor : Dr. Carenini
A well-written text is not merely a sequence of independent and isolated sentences, but instead a sequence of structured and related sentences. It addresses a particular topic, often covering multiple subtopics, and is organized in a coherent way that enables the reader to process the information. Discourse analysis seeks to uncover such underlying structures, which can support many applications including text summarization and information extraction. This thesis focuses on building novel computational models of different discourse analysis tasks in asynchronous conversations; i.e., conversations where participants communicate with each other at different times (e.g., emails, blogs). Effective processing of these conversations can be of great strategic value for both organizations and individuals. We propose novel computational models for topic segmentation and labeling, rhetorical parsing and dialog act recognition in asynchronous conversation. Our approaches rely on two related computational methodologies: graph theory and probabilistic graphical models. The topic segmentation and labeling models find the high-level discourse structure; i.e., the global topical structure of an asynchronous conversation. Our graph-based approach extends state-of-the-art methods by integrating a fine-grained conversational structure with other conversational features. On the other hand, the rhetorical parser captures the coherence structure, a finer discourse structure, by identifying coherence relations between the discourse units within each comment of the conversation. Our parser applies an optimal parsing algorithm to probabilities inferred from a discriminative graphical model which allows us to represent the structure and the label of a discourse tree constituent jointly, and to capture the sequential and hierarchical dependencies between the constituents. Finally, the dialog act model allows us to uncover the underlying dialog structure of the conversation. We present unsupervised probabilistic graphical models that capture the sequential dependencies between the acts, and show how these models can be trained more effectively based on the fine-grained conversational structure. Together, these structures provide a deep understanding of an asynchronous conversation that can be exploited in the above-mentioned applications. For each discourse processing task, we evaluate our approach on different datasets, and show that our models consistently outperform the state-of-the-art by a wide margin. Often our results are highly correlated with human annotations.
Periodic vibrotactile guidance
Karuei, Idin
DOI : 10.14288/1.0167014
URI : http://hdl.handle.net/2429/50772
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-11
Supervisor : Dr. MacLean
Emergence of mobile technologies, with their ever increasing computing power, embedded sensors, and connectivity to the Internet has created many new applications such as navigational guidance systems. Unfortunately, these devices can become problematic by inappropriate usage or overloading of the audiovisual channels. Wearable haptics has come to the rescue with the promise of offloading some of the communication from the audiovisual channels. The main goal of our research is to develop a spatiotemporal guidance system based on the potentials and limitations of the sense of touch. Our proposed guidance method, Periodic Vibrotactile Guidance (PVG), guides movement frequency through periodic vibrations to help the user achieve a desired speed and/or finish a task in a desired time. We identify three requirements for a successful PVG system: accurate measurement of the user's movement frequency, successful delivery of vibrotactile cues, and the user's ability to follow the cues at different rates and during auditory multitasking. In Phase 1, we study the sensitivity of different body locations to vibrotactile cues with/without visual workload and under different movement conditions and examine the effect of expectation of location and gender differences. We create a set of design guidelines for wearable haptics. In Phase 2, we develop Robust Realtime Algorithm for Cadence Estimation (RRACE) which measures momentary step frequency/interval via frequency-domain analysis of accelerometer signals available in smartphones. Our results show that, with a 95% accuracy, RRACE is more accurate than the published state-of-the-art time-based algorithm. In Phase 3, we use the guidelines from Phase 1 and the RRACE algorithm to study PVG. First we examine walkers' susceptibility to PVG which shows most walkers can follow the cues with 95% accuracy. Then we examine the effect of auditory multitasking on users' performance and workload, which shows that PVG can successfully guide the walker's speed during multitasking. Our research expands the reach of wearable haptics and guidance technologies by providing design guidelines, a robust cadence detection algorithm, and Periodic Vibrotactile Guidance -- an intuitive method of communicating spatiotemporal information in a continuous manner -- which can successfully guide movement speed with little to no learning required.
Partitioning and distribution of web applications to the hybrid cloud
Kaviani, Nima
DOI : 10.14288/1.0165945
URI : http://hdl.handle.net/2429/50017
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-11
Supervisor : Dr. Wohlstadter
Hybrid cloud deployment is an effective strategy in deploying software services across public cloud and private infrastructure. It allows deployed software systems to benefit from cost savings and scalability offerings of the cloud while keeping control over privacy- or security-sensitive code and data entities. However, the complexity of determining which code and data entities should reside on-premises, and which can be migrated to the cloud is daunting. Researchers have attempted to address this complexity by using partitioning algorithms to optimize distribution and deployment of code entities across public cloud and private infrastructure. However, we have identified the following shortfalls with the existing research work: Current research does not provide enough flexibility in placement of software function execution and data entities between public/private hosts. In particular it does not allow for replication or optimized separation of code and data entities in relation to one another. Current research on partitioning of software systems does not explicitly consider the dynamics of a hybrid cloud deployment when making decisions about public cloud and private infrastructure. Particularly, current research lacks support for making explicit tradeoff s between monetary cost and improved performance in hybrid cloud software systems. The dynamics of the cloud require partitioning algorithms to be tailored towards features inherent to a hybrid cloud deployment. This includes encoding data dependency models and component dependency models of a software system collectively into one unique mathematical optimization model. There is no existing algorithm that allows for combined code and data dependency requirements to be modelled under one optimization formula. This thesis presents my work on implementing algorithms and tools that address the shortcomings of the previous research as discussed above. These algorithms and tools are put together under a partitioning and distribution framework named Manticore. Manticore has been used to drive partitioning and deployment decisions on several open source software systems. The experiment results show an estimate of up to 54% reduction in monetary costs compared to a premises only deployment and 56% improvement in performance compared to a na ive separation of code entities from data entities in a hybrid cloud deployment.
Relational logistic regression
Kazemi, Seyed Mehran
DOI : 10.14288/1.0166004
URI : http://hdl.handle.net/2429/50091
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Poole
Multipresenter++: A Presentation System for Multiple Display Screens
Li, Huan
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Booth
Reinforcement learning using sensorimotor traces
Li, Jingxian
DOI : 10.14288/1.0103383
URI : http://hdl.handle.net/2429/45590
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. van de Panne
A model and analysis of two-handed interaction with a keyboard and pointing device
Link, Juliette Frances
DOI : 10.14288/1.0052171
URI : http://hdl.handle.net/2429/45417
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisors : Dr. McGrenere, Dr. Booth
[no title]
Mahmud, Anil
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Woodham
An empirical study of practical, theoretical and online variants of random forests
Matheson, David James
DOI : 10.14288/1.0167214
URI : http://hdl.handle.net/2429/46586
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. de Freitas
Object classification and localization with spatially localized features
McCann, Sancho Juan Carlos Dean Rob Roy
DOI : 10.14288/1.0167312
URI : http://hdl.handle.net/2429/46525
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-05
Supervisor : Dr. Lowe
Object classification and localization are important components of image understanding. For a computer to interact with our world, it will need to identify the objects in our world. At a more basic level, these tasks are crucial to many practical applications: image organization, visual search, autonomous vehicles, and surveillance. This thesis presents alternatives to the currently popular approaches to object classification and localization, specifically focusing on methods that more tightly integrate location information with the visual features. We start by improving on Naive Bayes Nearest Neighbor (NBNN), an alternative to the standard bag-of-words/spatial pyramid classification pipeline. This model matches localized features between a test image and the entire training set in order to classify an image as belonging to one of several categories. We improve this method’s classification performance and algorithmic complexity. However, the nature of NBNN results in prohibitive memory requirements in large datasets. This leads to our second contribution: a bag-of-words model based on a clustering of the location-augmented features. This a simple and more flexible approach to modeling location information than the commonly used spatial pyramid. By using location-augmented features, location information is captured simply in the nearest-neighbor coding of the bag-of-words model. This results in a more efficient use of model dimensions than the spatial pyramid and higher classification performance than state-of-the-art alternatives. Last, we present the design of an object localization system using this high performance classifier. Such design is made more difficult by the fact that our model does not satisfy the assumptions made by recent efficient localization algorithms. Our method uses a Hough transform based on an approximation to our model, followed by a more accurate refinement of classifier scores and bounding boxes. We show its effectiveness on the widely used and practical Daimler monocular pedestrian dataset. These contributions show that simple, location-augmented features, soft-nearest neighbor coding, and linear Support Vector Machines (SVMS) can outperform the long-used and optimized spatial pyramid methods and that this approach warrants additional research to continue to improve its efficiency.
Towards human pose estimation in video sequences
Oleinikov, Georgii
DOI : 10.14288/1.0166854
URI : http://hdl.handle.net/2429/45767
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Little
Scrolling in radiology image stacks : multimodal annotations and diversifying control mobility
Oram, Louise Carolyn
DOI : 10.14288/1.0165660
URI : http://hdl.handle.net/2429/45508
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. MacLean, Dr. Kruchten (Electrical and Computer Engineering)
Automatic abstractive summarization of meeting conversations
Oya, Tatsuro
DOI : 10.14288/1.0165907
URI : http://hdl.handle.net/2429/49946
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Carenini
System support for elasticity and high availability
Rajagopalan, Shriram
DOI : 10.14288/1.0166887
URI : http://hdl.handle.net/2429/46272
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-05
Supervisor : Dr. Warfield
Elasticity and high availability (HA) are key requirements among modern Internet applications. Elasticity enables applications to dynamically allocate/release physical resources in proportion to request load. High availability enables applications to mask failures in the system from end users. In current practice, every application implements these features as part of its own application logic, resulting in unnecessary design complexity. This thesis argues that facilities for elasticity and HA should be exposed as system-level primitives, in the same way abstractions for files and networks became operating system-level primitives three decades ago. Unfortunately, providing these higher-level services efficiently may require knowledge of application data structures, consistency requirements, and workloads. This thesis describes initial instantiations of such interfaces for two broad (and different) classes of applications: network middleboxes (e.g., load balancers, intrusion prevention systems, etc) and database systems. Elasticity is achieved typically through dynamic partitioning of state and inputs into independent subsets, while HA is achieved through state replication. Guided by this principle, this thesis presents a system-level runtime that partitions the middlebox state along flow boundaries and provides abstractions for elasticity and HA using live migration and replication of flows respectively. For database systems, this thesis presents a hypervisor-level HA system that performs database-aware virtual machine replication, eliminating the need for complex application-level HA mechanisms. This thesis concludes that while there may not be a one-size-fits-all solution to application elasticity and HA, it is still feasible and beneficial to provide system-level primitives that are applicable across one or more application domains.
Stability in markets with power asymmetry
Rastegari, Baharak
DOI : 10.14288/1.0072135
URI : http://hdl.handle.net/2429/45695
Degree : Doctor of Philosophy - PhD
Graduation Date : 2014-05
Supervisors : Dr. Condon, Dr. Leyton-Brown
Two classes of widely studied markets are auctions, such as eBay, and two-sided matching markets, such as matching medical residents to hospitals. In both of these markets, it often happens that one side is in power and can influence the outcome of the market---e.g., the seller in an auction. In the fist part of this thesis we study two-sided matching markets in which participants are initially endowed with partial preference orderings. Our goal is to identify a matching that is stable and optimal for the side of the market that is in power, w.r.t. the underlying true preferences of the agents. We first investigate the extent to which we can learn about this matching given the partial information. We then provide a novel model in which the true preferences are learned through interviews. Our goal is to identify a centralized interviewing policy that returns our matching of interest while minimizing the number of interviews. We introduce three minimization criteria, of which the most desirable one is not always achievable. We provide an exponential-time algorithm for computing a policy that satisfies the other two criteria, and exhibit evidence that a more efficient computation may not be possible in general. We then show how to design a computationally efficient interview-minimizing policy for a setting in which agents on one side of the market are initially endowed with identical partial preferences, and agents must interview before getting matched. In the second part, we study combinatorial auctions (CAs), where multiple goods are for sale and buyers are allowed to place bids on bundles, i.e. sets of goods. In single-good auctions the auctioneer is at no risk of losing revenue if more bidders compete in the auction. In CAs however, an auctioneer may fi nd it pro table to disqualify bidders in order to be matched to a smaller set of bidders. We investigate the extent to which this counterintuitive phenomenon can occur under CA mechanisms that o er bidders dominant strategies. We show that a broad class of deterministic CAs exhibit non-monotonicity in revenue. However, revenue monotonic CAs exist in the broader domain of randomized mechanisms.
Near-optimal Herding
Samadi, Samira
DOI : 10.14288/1.0166033
URI : http://hdl.handle.net/2429/50167
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Harvey
Predicting parameters in deep learning
Shakibi, Babak
DOI : 10.14288/1.0165555
URI : http://hdl.handle.net/2429/50999
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. de Freitas
Design and optimization of control primitives for simulated characters
Shen, Shuo
DOI : 10.14288/1.0165691
URI : http://hdl.handle.net/2429/45617
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. van de Panne
Community detection in sparse time-varying networks
Slind, Jillian Rae
DOI : 10.14288/1.0165964
URI : http://hdl.handle.net/2429/50043
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Ng
Modeling ordinal data for recommendation system
Srivastava, Anupam
DOI : 10.14288/1.0166985
URI : http://hdl.handle.net/2429/50433
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Poole
A stable and robust method to identify modules of functionally coherent genes
Takhar, Mandeep Kaur
DOI : 10.14288/1.0166054
URI : http://hdl.handle.net/2429/50675
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Ng
A risk assessment infrastructure for powered wheelchair motion commands without full sensor coverage
TalebiFard, Pouria
DOI : 10.14288/1.0103390
URI : http://hdl.handle.net/2429/46276
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Mitchell
Auto-WEKA : combined selection and hyperparameter optimization of supervised machine learning algorithms
Thornton, Christopher
DOI : 10.14288/1.0165896
URI : http://hdl.handle.net/2429/46177
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisors : Dr. Leyton-Brown, Dr. Hoos
Managing updates and transformations in data sharing systems
Thrastarson, Arni Mar
DOI : 10.14288/1.0135578
URI : http://hdl.handle.net/2429/50729
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Pottinger
Real-time predictions from unlabeled high-dimensional sensory data during non-prehensile manipulation
Troniak, Daniel Michael
DOI : 10.14288/1.0167630
URI : http://hdl.handle.net/2429/50910
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisors : Dr. Pai, Dr. van de Panne
Conflict-driven symbolic execution : how to learn to get better
Val, Celina Gomes do
DOI : 10.14288/1.0165906
URI : http://hdl.handle.net/2429/46226
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. Hu
Towards a time-lapse prediction system for cricket matches
Veppur Sankaranarayanan, Vignesh
DOI : 10.14288/1.0167478
URI : http://hdl.handle.net/2429/46870
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Lakshmanan
Iterative solution of a mixed finite element discretisation of an incompressible magnetohydrodynamics problem
Wathen, Michael Philip
DOI : 10.14288/1.0135538
URI : http://hdl.handle.net/2429/50202
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisors : Dr. Greif, Dr. Schoetzau (Mathematics)
Formal verification of a digital PLL
Wei, Jijie
DOI : 10.14288/1.0165974
URI : http://hdl.handle.net/2429/50048
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisor : Dr. Greenstreet
Binary shuffling : defeating memory disclosure attacks through re-randomization
Williams-King, David Christopher
DOI : 10.14288/1.0167580
URI : http://hdl.handle.net/2429/48600
Degree : Master of Science - MSc
Graduation Date : 2014-11
Supervisors : Dr. Aiello, Dr. Warfield
Exploring structured predictions from sensorimotor data during non-prehensile manipulation using both simulations and robots
Zhu, Chuan
DOI : 10.14288/1.0165671
URI : http://hdl.handle.net/2429/45552
Degree : Master of Science - MSc
Graduation Date : 2014-05
Supervisor : Dr. van de Panne