Search Constraints
Filtering by:
Language
English
Remove constraint Language: English
Publisher
Carleton University
Remove constraint Publisher: Carleton University
Number of results to display per page
Search Results
-
- Resource Type:
- Thesis
- Creator:
- Askin, Enver Deniz
- Abstract:
- Transformers are machine learning models designed to learn and predict sequential and structured data, which are crucial to tasks such as neural machine translation and semantic parsing. They have become state-of-the-art engines for both of these tasks and much research in natural language processing is devoted to increasing their performance by introducing modifications to their architectures. In light of this trend, this thesis introduces a new Transformer architecture called MAWT: Multi-Attention-Weight Transformers in an attempt to increase the accuracy and variety of the acceptable predictions of a Transformer. It attempts to achieve this by training multiple weights per each Transformer attention head, which then are used to test the accuracy of the engine. This creates a new architecture under which the system produces a candidate set of outputs (instead of a singly output), along with a method for selecting from the candidate set. My proposal rests on the assumption -- motivated by statistical considerations -- that having a candidate set increases the probability of finding an exact match within the set. Upon testing, I observed that my system outperforms the regular transformer on 5/6 benchmark neural machine translation and semantic parsing datasets, where engine performance is measured by exact match accuracy. Exact match accuracy demands syntactic identity between the output and the target. In order to investigate how well my new architecture generalizes to measures of semantic equivalence that don't also demand syntactic identity, I also recorded the BLEU scores on these datasets. The BLEU score is a measure of performance based on n-grams rather than exact symbolic match (i.e., how many contiguous sequence of n-many strings from the predicted output match the desired output). The results I report on the BLEU scores are more mixed, raising important questions that I highlight about the role of syntax in measures of semantic equivalence.
- Thesis Degree:
- Doctor of Philosophy (Ph.D.)
- Thesis Degree Discipline:
- Cognitive Science
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Sheedy, Krysten Lynn
- Abstract:
- Glufosinate is the second most commonly used herbicide worldwide; it inhibits glutamine synthetase, which results in increased ammonia levels in plants and mammals. Due to their high polarity, low volatility, small size and lack of chromophores and fluorophores, glufosinate and its breakdown product 3-(methylphosphinico) propionic acid (3-MPPA) are difficult to detect at trace levels. Using the chemical derivatization strategy trimethylation enhancement using diazomethane (TrEnDi), glufosinate and 3-MPPA react with diazomethane and tetrafluoroboric acid to become permethylated, thus reducing their polarity and forming a fixed permanent positive charge on the amine group of glufosinate. When using reversed-phase high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS), analyte retention and sensitivity are increased after derivatization, 6.7-fold for glufosinate and 6.3-fold for 3-MPPA. TrEnDi methodology was applied to canola samples from two separate fields sprayed with Liberty®. Prior to derivatization, the analysis showed no signal associated with unmodified glufosinate or 3- MPPA; however, TrEnDi modification resulted in quantifiable signals for both permethylated species.
- Thesis Degree:
- Master of Science (M.Sc.)
- Thesis Degree Discipline:
- Chemistry
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Bernyk, Kimberly Ellen
- Abstract:
- Remotely piloted aircraft systems are a remote sensing platform for peatland researchers to monitor changes in vegetation height/composition/structure. In this work, an RPAS was flown to collect aerial images over Alfred Bog, a domed peatland complex located in Eastern Ontario. The images were processed with a photogrammetry technique called Structure-from-Motion, which can be used to create 3D point clouds of an x by y transect. The point cloud results were used to assess the utility of extracting ground terrain and vegetation height as compared to a transect field survey. This field survey was completed with a Trimble Catalyst RTK GNSS to record ground elevation and maximum vegetation height of the canopy top. The results from this research suggest that terrain information could not be extracted at all from the generated point clouds. Although, a digital surface model can be generated to model the canopy top and crown area.
- Thesis Degree:
- Master of Science (M.Sc.)
- Thesis Degree Discipline:
- Geography
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Naranjo Vera, Mairelys
- Abstract:
- The wood frog, Rana sylvatica, is well known for its freeze tolerance ability. To endure winter, frozen frogs switch to a hypometabolic state via transcriptional regulation. Histone methylation is known to play a crucial role in regulating gene transcription. However, histone arginine methylation or demethylation has not previously been studied in the context of freeze tolerance. This thesis presents the first characterization of arginine methylation in a freeze tolerant vertebrate. Overall, levels of protein arginine methyltransferases (PRMTs) and methylated histone residues showed differential regulation over the freeze/thaw-cycle in wood frog liver. All PRMTs and downstream targets showed no changes during freezing, but protein levels of targets associated with transcription activation were elevated during thaw in skeletal muscle. Differential levels of histone demethylases were found in both tissues among the experimental conditions. These results indicate a role for histone methylation in supporting metabolic rate depression and tissue homeostasis during freezing.
- Thesis Degree:
- Master of Science (M.Sc.)
- Thesis Degree Discipline:
- Biology
- Date Created:
- 2022
-
- Resource Type:
- Thesis
- Creator:
- Xia, Xuexin
- Abstract:
- Mining is a booming industry and it indirectly involves everyone living in modern society. Tailings, the primary solid waste generated as the side effect of extracting valuable minerals, often require vast facilities to store the large quantity of waste. Failure of tailings storage facilities often causes significant damage both environmentally and economically, and frequently results in fatalities. Nowadays numerical simulation of tailings flows resulting from potential failures has become widespread in practice to assist the design of tailings storage facilities. In this thesis, tailings runout simulations are attempted using a numerical method suitable for large deformation analysis (Material Points Method) employing an advanced rheological model. This study aims to simulate the runout of tailings dam breach incident such as Merriespruit in South Africa using realistic geotechnical properties and producing results that fulfill the expectations from both geo-mechanical and hydrodynamical requirements.
- Thesis Degree:
- Master of Applied Science (M.App.Sc.)
- Thesis Degree Discipline:
- Engineering, Environmental
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Polakova, Laura
- Abstract:
- The study examined the impact of the COVID-19 pandemic on treatment access among individuals (n=225) presenting with problematic opioid use to the Rapid Access Addiction Medicine (RAAM) clinic at The Royal in Ottawa, Ontario. The COVID-19 pandemic led to government-imposed restrictions and service limitations. The RAAM clinic underwent two primary changes: (1) delivering services virtually rather than in-person appointments and (2) shifting from walk-in services to appointment-based services. The study was a retrospective chart review and data were extracted from an electronic health record, Meditech. Participants were patients who had an initial presentation to the RAAM clinic between March 16th, 2019, and March 15th, 2021, and had used opioids within the 30 days prior to their visit. Results indicated that RAAM changes decreased some groups' access to care such that fewer patients experiencing precarious housing and mental health comorbidities presented to the RAAM clinic after the onset of the pandemic.
- Thesis Degree:
- Master of Arts (M.A.)
- Thesis Degree Discipline:
- Psychology
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Augustin, Claudia Tharis
- Abstract:
- The Americano do Brasil and Mangabal Complexes belong to a cluster of Neoproterozoic mafic-ultramafic intrusions in the southern Goiás Magmatic Arc in Brazil that lies within the Brasília Belt, a complex, long-lived Neoproterozoic Orogen that formed during the amalgamation of the São Francisco/Congo, Amazonian, and Paranapanema Cratons in western Gondwana. Although deformed and partially recrystallized by a regional metamorphic overprint, the rocks of both complexes still preserve igneous textures. The rocks in both complexes show cumulate textures that range from adcumulate to orthocumulate. Petrography, geochemistry, geochronology, and petrological modeling were carried out to constrain the petrogenesis of the complexes. Modeling indicates that the parental magma for the Americano do Brasil Complex had an OIB-like Fe-rich picrite composition, while the parental magma of the Mangabal Complex could have been a tholeiitic basalt. Considering the temporal and spatial correlations, decompression melting of upwelling asthenosphere triggered by rollback subduction and/or influence of a mantle plume are plausible mechanisms to explain the petrogenesis of these mafic-ultramafic intrusions in the vicinities of the eastern Arenópolis arc and westernmost Anápolis-Itauçu belt. The complexes have experienced a protracted metamorphic evolution with peaks indicating amphibolite facies. The rock from both complexes shows late-stage alteration associated with metasomatism by CO2-Ca-S-rich fluids.
- Thesis Degree:
- Doctor of Philosophy (Ph.D.)
- Thesis Degree Discipline:
- Earth Sciences
- Date Created:
- 2022
-
- Resource Type:
- Thesis
- Creator:
- Saffarzadeh Parizi, Sorousha
- Abstract:
- This thesis evaluates the safety impacts of red-light cameras (RLCs) and Dynamic Speed Display Signs (DSDSs) in Ottawa, Canada. The study examines the safety impacts of RLCs on safety performance, using collision records, and driver behaviour using surrogate safety measures. The safety impacts of DSDSs on driver behaviour are evaluated using speed analysis. An empirical Bayes method for RLCs showed a significant impact, where total and PDO collisions increased while injury and fatal collisions decreased. The impact of RLCs also depended on the collision types, where sideswipe, rear-end, and SMV collisions increased, but the angle and turning collisions decreased. The increase in rear-end collisions was also examined through an analysis of traffic conflicts. The results indicated that treated sites had significantly more severe rear-end conflicts that had likely resulted from harder deceleration rates. Speed analysis for DSDSs indicated drivers reduced their speed when they saw their actual speed on DSDSs.
- Thesis Degree:
- Master of Applied Science (M.App.Sc.)
- Thesis Degree Discipline:
- Engineering, Civil
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Costa, Alexsander Cassio Aguiar Antunes
- Abstract:
- In this thesis work, the use of UAS in the study of migratory shorebird species in Canada is explored with the development of computer vision applications. A deep learning classification model is trained to identify the presence of birds of a given species in an image. Images were collected from UAS for the development of the vision models, and realistic models of the species of interest were used. To address a data scarcity issue, the datasets used were augmented with synthetic data with realistic models of the birds. For evaluation of the quality of the artificially generated images, a novel measure is developed. The synthetic image quality measure showed better results in controlled environments when compared to a popular alternative in the literature. The classifiers trained with the augmented dataset showed appropriate performance, with mean accuracy and standard deviation of 94% +- 0.04 in the test set.
- Thesis Degree:
- Master of Applied Science (M.App.Sc.)
- Thesis Degree Discipline:
- Engineering, Mechanical
- Date Created:
- 2023
-
- Resource Type:
- Thesis
- Creator:
- Petropoulos, Amanda
- Abstract:
- Intermediate care units have been proposed to meet the needs of incarcerated individuals with moderate mental health needs. Few studies have examined their profiles, and nature and effectiveness of interventions received. In the current study, the profiles of 295 incarcerated men in Canadian federal institutions were examined. File reviews were conducted over a 12 month period to extract treatment information as well as assess changes in number of incidents of self-harm, attempted suicide, overdose, and violence, and mental health needs. Coarsened exact matching and descriptive analyses revealed differences in the profiles across three treatment levels (mental health care delivered in: intermediate care units, treatment centres or mainstream institutions); roughly 30% improvement on all outcomes was observed across the groups. Overall findings highlight the need to further explore how best to meet the needs of incarcerated individuals assessed with moderate mental health needs, as this remains important for future management.
- Thesis Degree:
- Master of Arts (M.A.)
- Thesis Degree Discipline:
- Psychology
- Date Created:
- 2023