Not available / text

Not available / text

This thesis addresses the problem of accounting for morphological alternation within Natural Language Processing. It proposes an approach to morphology which is based on phonological concepts, in particular the syllable, in contrast to morpheme-based approaches which have standardly been used by both NLP and linguistics. It is argued that morpheme-based approaches, within both linguistics and NLP, grew out of the apparently purely affixational morphology of European languages, and especially English, but are less appropriate for non-affixational languages such as Arabic. Indeed, it is claimed that even accounts of those European languages miss important linguistic generalizations by ignoring more phonologically based alternations, such as umlaut in German and ablaut in English. To justify this approach, we present a wide range of data from languages as diverse as German and Rotuman. A formal language, MOLUSe, is described, which allows for the definition of declarative mappings between syllable-sequences, and accounts of non-trivial fragments of the inflectional morphology of English, Arabic and Sanskrit are presented, to demonstrate the capabilities of the language. A semantics for the language is defined, and the implementation of an interpreter is described. The thesis discusses theoretical (linguistic) issues, as well as implementational issues involved in the incorporation of MOLUSC into a larger lexicon system. The approach is contrasted with previous work in computational morphology, in particular finite-state morphology, and its relation to other work in the fields of morphology and phonology is also discussed.

published_or_final_version / abstract / toc / Computer Science / Doctoral / Doctor of Philosophy

No description available.

Computational imaging (CI) systems are hybrid imagers in which the optical and post-processing sub-systems are jointly optimized to maximize the task-specific performance. In this dissertation we consider a form of CI system that measures the linear projections (i.e., features) of the scene optically, and it is commonly referred to as feature-specific imaging (FSI). Most of the previous work on FSI has been concerned with image reconstruction. Previous FSI techniques have also been non-adaptive and restricted to the use of ambient illumination.We consider two novel extensions of the FSI system in this work. We first present an adaptive feature-specific imaging (AFSI) system and consider its application to a face-recognition task. The proposed system makes use of previous measurements to adapt the projection basis at each step. We present both statistical and information-theoretic adaptation mechanisms for the AFSI system. The sequential hypothesis testing framework is used to determine the number of measurements required for achieving a specified misclassification probability. We demonstrate that AFSI system requires significantly fewer measurements than static-FSI (SFSI) and conventional imaging at low signal-to-noise ratio (SNR). We also show a trade-off, in terms of average detection time, between measurement SNR and adaptation advantage. Experimental results validating the AFSI system are presented.Next we present a FSI system based on the use of structured light. Feature measurements are obtained by projecting spatially structured illumination onto an object and collecting all of the reflected light onto a single photodetector. We refer to this system as feature-specific structured imaging (FSSI). Principal component features are used to define the illumination patterns. The optimal LMMSE operator is used to generate object estimates from the measurements. We demonstrate that this new imaging approach reduces imager complexity and provides improved image quality in high noise environments. We then generalize the FSSI system by making use of random projections (i.e., using no object prior) to define the illumination patterns. Object estimates are generated using L1-norm minimization and gradient-projection sparse reconstruction algorithms. Experimental results validating the FSSI system are presented.

Heart development has been extensively studied in numerous organisms throughout the twentieth century. The timing of key inductive signals and the expression of many critical transcription factors have been mapped across a variety of model systems. A collective image of the various stages of cardiac development is beginning to emerge. Although most of the seminal events are conserved across evolution, it is increasingly clear that subtle differences can have substantive effects on models of heart development processes. Furthermore, the overwhelming majority of work contributing to these models has been performed on a gene-by-gene basis. As a result, we have a loosely stitched cross-evolutionary view of cardiogenesis that leaves much to be desired by way of completeness. Thus, in order to move toward a comprehensive model of heart development, we have a critical need for global network views of heart development processes conducted within one species.Cardiac myogenesis, the development of heart muscle cells, is the earliest heart development process and is required for the formation of all adult heart structures. Key signaling pathways, and their precise timing and targets, have only recently begun to be defined. The downstream targets of these pathways and their timing of activation or repression remain largely unknown. To address this, I compiled data from three genomic microarray studies, each addressing a distinct aspect of cardiac myogenesis signaling and expression, to construct a global preliminary network of the primary inductive signals and their downstream targets in the chick model embryology system.The preliminary cardiac myogenesis network obtained from these studies generates far too many hypotheses to test experimentally. The challenge that lies ahead for elucidating the fine structure of this, or any network model, is in determining the next most enlightening experiments. Headway in sorting out more profitable experiments can be made by selecting from among the universe of known interaction data as well as taking advantage of a property selected for throughout evolution - robustness. Network robustness is loosely defined as the ability of a network to maintain input and output properties in the face of perturbation. It is unsurprising that evolution would sculpt such a characteristic into molecular networks required to perform a task in varied environmental and genetic circumstances. However the way in which evolution has engendered this quality has opened the door to an exciting new avenue for in silico experimentation.I present in this dissertation the beginnings of a collaborative project for biological network elucidation software called BioNET. The long-term goal of BioNET is to take a description of a network model and phenotype as input and return a set of candidate network models capable of more robustly producing the phenotype. Fundamental to BioNET is the ability to acquire information from the universe of known molecular interaction data for in silico experimentation in any model system. To this end, I redesigned BioNetBuilder, open-source network integration software, to transfer any and all publicly available interaction data across species and serve them via the web. As these data grow in scale, BioNET will be increasingly useful for identifying the more plausible, among possible network architectures, such as the preliminary cardiac myogenesis network presented in this dissertation.

This thesis explores the intersection between computation and architectural design. The thesis first develops several computational design tools, specifically focussing on three problem domains: (1) speedy generation and modification of architectural schemes sharing a common typological language, (2) analysis of urban and neighbourhood conditions, and (3) performance modelling and prediction
. To test the tools, the thesis subsequently applies the tools to design several variations of a condominium tower in downtown Toronto. Despite some limitations, the computational toolkit proved powerful and flexible enough to generate viable condominium schemes under various sets of assumptions.

Graphene is a single atomic layer of graphite and has many extraordinary properties. Many graphene based applications have been proposed in recent years and the need of a time domain simulation tool for studying graphene based devices emerges. This thesis focuses on developing a simulation framework for graphene based devices using finite-difference time-domain (FDTD) method. Formulation for a perfectly matched layer (PML) for the sub-cell FDTD method for thin dispersive layers has been derived and implemented. Such a PML is useful when thin layers extend to the boundaries of the computational domain. Using the sub-cell PML formulation to model the graphene thin layers significantly reduces the computational cost compared to using the conventional FDTD. The proposed formulation is accompanied by detailed validation and error analysis studies. Several graphene applications are simulated using the new framework and the results show good agreement with the respective analytical models.

Graphene is a single atomic layer of graphite and has many extraordinary properties. Many graphene based applications have been proposed in recent years and the need of a time domain simulation tool for studying graphene based devices emerges. This thesis focuses on developing a simulation framework for graphene based devices using finite-difference time-domain (FDTD) method. Formulation for a perfectly matched layer (PML) for the sub-cell FDTD method for thin dispersive layers has been derived and implemented. Such a PML is useful when thin layers extend to the boundaries of the computational domain. Using the sub-cell PML formulation to model the graphene thin layers significantly reduces the computational cost compared to using the conventional FDTD. The proposed formulation is accompanied by detailed validation and error analysis studies. Several graphene applications are simulated using the new framework and the results show good agreement with the respective analytical models.