39568 research outputs found
Sort by
Local Catch Canada Retreat: Summary Report
No full textThis report presents a summary of the Local Catch Canada Network 2025 Retreat, including key outputs of network workshops and action items to guide the network forward.The Oceans Collaborativ
MODELLING PHYTOPLANKTON GROWTH RATES UNDER CHANGING RESOURCE CONDITIONS
No full textPhytoplankton are central to global biogeochemical cycles and marine ecosystems, yet accurately modeling their growth and photosynthetic responses under dynamic environmental conditions remains a major challenge. This thesis addresses fundamental questions in this domain by introducing original models that explain microbial growth dynamics under abrupt nutrient shifts using minimal state variables, and by developing a parsimonious framework for interpreting photosynthesis–irradiance responses. These ideas are grounded in advanced mathematical concepts, including fractional calculus and set theory, yet practical and easy to apply to real data. Their validity is rigorously tested using a suite of statistical methods—including non-linear optimization, linear and non-linear regression, generalized additive mixed models, and machine learning methods—applied to extensive laboratory and open-ocean datasets.
Chapter 1 introduces the concept of physiological memory into microbial growth model through the Monod-memory model. Grounded in robust mathematical theory, this model quantifies memory as a biologically meaningful parameter measurable in laboratory settings. By integrating the computational efficiency of the classical Monod model with the physiological depth of the Droop model, it adequately captures population dynamics under nutrient-limited and starvation conditions. In addition to nutrients, light availability is another key driver of phytoplankton growth, typically characterized by the photosynthesis–irradiance (P–I) curve. Chapter 2 focuses on this relationship, introducing a novel P–I model that accurately captures the plateau part of the curve and decline in photosynthesis rate associated with photoinhibition—an area where traditional models have struggled for over a century. This formulation enhances both the fit to empirical data and the interpretability of P–I parameters, marking a significant advancement in light-response modeling. Chapter 3 extends this work by introducing a hybrid statistical–physiological framework that links P–I parameters to environmental drivers, enabling large-scale, mechanistic interpretations of phytoplankton photosynthesis. Chapter 4 presents two independent open-source R packages developed as part of this thesis, promoting transparency, reproducibility, and broader adoption of the models. Collectively, these contributions bridge theoretical innovation with empirical rigor, significantly advancing the modeling of phytoplankton growth and function in dynamic environmental contexts
Evaluating Few Shot Learning With Uncertainty Quantification Under Encrypted Traffic Classification
No full textN/AComputer networks frequently contain novel applications which a network administrator could benefit from understanding. Encryption and virtual-private-network (VPN) service providers are at odds with this. This thesis contains a benchmark and study of the Out-Of-Distribution (OOD) detection and classification of encrypted VPN traffic with few shot learning algorithms. The research determines the impactful hyperparameters of a few shot learning algorithms under the VPN setting. The work then moves to better understand the OOD detection performance, testing alternative few shot learners and finding potential trade offs between them. The research finds that a transductive, few shot learner has superior OOD detection to its inductive counterpart. However, transductive methods typically require more data to configure. Therefore, the research develops and tests a hybrid inductive-transductive approach, thus determining if a middle ground is possible without too many negative consequences
Living Scaffold: A Framework for Multispecies Architecture
No full textThe Sandy Lake area of Bedford, Nova Scotia, serves as a vital wildlife corridor, and its health is essential to the region’s ecology. Increasing housing demands, urban sprawl, and its proximity to urban areas render it particularly vulnerable to development. Following a 300-acre clear-cut in 2013, residents began advocating for an expansion of the area’s protected boundaries.
This thesis explores how architecture can engage with local ecologies and wildlife as an alternative to conventional conservation methods that minimize human activity and development. Through the design of an ecological learning centre at Sandy Lake, architecture is positioned as a tool to support local biodiversity while encouraging humans to take a more active role in conservation. A living scaffold that benefits both humans and non-human entities contributes to the program’s educational agenda, setting the stage for a citizen science laboratory that fosters the connection between humans and nature
RAPPORT COMMUNAUTAIRE: Accés aux programmes de traitement de la toxicomanie financés par le gouvernement au Canada atlantique: perspectives des familles de personnes qui consomment des substances, du personnel des organismes communautaires et des directeurs/médecins des programmes de traitement (Résultats des phases 2 et 3 de l'étude COAST
No full text
Advanced analysis of contrast agents in X-ray and magnetic resonance imaging
No full textMedical imaging techniques, such as x-ray imaging and magnetic resonance imaging (MRI), allow us to visualize anatomical structures of the body and in some cases, study underlying biological processes. Contrast agents used in imaging aid in visualizing different structures and processes by increasing contrast differences, either generally or in a targeted fashion. In x-ray imaging, contrast agents enhance radiodensity in target tissues. In MRI, contrast agents work by shortening relaxation times of nuclei in the body. In both cases, contrast agents can yield important information about the mechanisms of disease and/or therapies and how best to optimize care.Medical imaging techniques, such as x-ray imaging and magnetic resonance imaging (MRI), allow us to visualize anatomical structures of the body and in some cases, study underlying biological processes. Contrast agents used in imaging aid in visualizing different structures and processes by increasing contrast differences, either generally or in a targeted fashion. In x-ray imaging, contrast agents enhance radiodensity in target tissues. In MRI, contrast agents work by shortening relaxation times of nuclei in the body. In both cases, contrast agents can yield important information about the mechanisms of disease and/or therapies and how best to optimize care.
For radiotherapy cancer treatment planning, the target area needs to be accurately delineated on a computed tomography (CT) image. In some instances, the target area may be difficult to see, resulting in inaccurate delineation of the target area. Materials modified with contrast agents can be placed in areas of interest after surgery to help accurately delineate the target. Molecular MRI is useful for assessing immunotherapy treatments for certain types of cancer. Contrast agents are used for cell labelling and in the case of brain tumours, assessing the structure of the blood brain barrier (BBB). We can image at different time points throughout treatment to monitor the immunotherapy treatment. With large amounts of imaging data, analysis becomes complex. We can apply a radiomics approach, which extracts features that are not obvious from individual images. Machine learning algorithms can be applied to determine if there are any correlations between features and treatment outcomes.
The objectives of this project were as follows. First was to modify a commercially available hydrogel material with a gadolinium-based contrast agent that can be injected into the surgical bed after lumpectomy so that it can be seen with CT, planar x-ray imaging and cone beam CT (CBCT). The second objective used radiomics and machine learning to identify potential preclinical MR imaging features that can be used as predictors for immunotherapy treatment success in a glioblastoma mouse model. Additionally, radiomics features were explored to determine if they could be used to differentiate between sex or treatment groups
ROLE OF LARGE TUMOR SUPPRESSOR PROTEINS (LATS1/2) IN NEURONAL MATURATION
No full textThe large tumor suppressor kinases (LATS1/2) are core kinases of the Hippo pathway. Little is known about their expression in the mature CNS. Here we show that Scribble can precipitate LATS2 protein from rat adult hippocampal lysate and that phosphorylated LATS (pLATS) and Scribble localize to the axon initiation segment (AIS) in hippocampal neurons. Using STED microscopy, we find that both Scribble and pLATS form distinct puncta with the Scribble puncta being190nm apart, similar to the pLATS puncta. Overexpression of myc-LATS2 and using a novel CRISPR dependent approach to label endogenous LATS2 with GFP, reveal that LATS2 is enriched in dendritic spines. Next, we developed a conditional approach to reduce LATS proteins in post mitotic neurons – this was unsuccessful. Nonetheless, this work identifies that LATS2 is expressed in hippocampal synapses, implicating it as an important kinase in learning and memory
Cooperative Urbanism: Cooperative Living at the Urban Scale
No full textMaster of Architecture ThesisThis thesis proposes a framework for approaching housing solutions through the lens of ‘cooperative urbanism,’ which expands the cooperative housing model to the urban scale, redirecting the focus from housing provision to the development of sustainable communities. The contemporary debate about urban housing stock and affordability continues to plague North American cities, remaining stalled in its search for a meaningful and lasting solution. Decision-makers at the municipal level persist in recycling familiar strategies, focusing on developer-led mid-rise and high-rise projects, with only sporadic attempts at social or cooperative housing models. Cooperative Urbanism offers an alternative by emphasizing shared spaces through the process of re-imagining traditional land-use patterns in urban areas. This thesis introduces the concept of “cooperative urbanism” in an architectural context to propose an alternate method of urban housing design on a redevelopment site in Halifax, Nova Scotia
LINEAR SHRINKAGE PRECISION MATRIX METHOD FOR IMPROVED FUNCTIONAL CONNECTIVITY ESTIMATION IN NEUROIMAGING
No full textNeurological disorders, arising from abnormalities in the nervous system, impose a major global health burden due to the lack of curative treatments and challenges in early diagnosis. Resting-state functional MRI (rs-fMRI) enables non-invasive mapping of brain activity, but conventional functional connectivity (FC) estimation often fails to capture complex network interactions. This study combines machine learning with shrinkage-based FC estimation to identify discriminative features that may serve as biomarkers across multiple disorders.
Five FC methods—Pearson’s, Spearman’s, Empirical Covariance, Ledoit–Wolf, and Oracle Approximating Shrinkage—were compared using classifiers including Logistic Regression, SVM, Random Forests, KNN, Naive Bayes, and CNNs. L1-regularized models guided feature selection.
Shrinkage-based estimators outperformed traditional methods, and a proposed Weighted Connectivity Matrix further improved accuracy, particularly with interpretable classifiers. Synthetic data confirmed robustness, and sensitivity analysis showed greatest influence from ROI count, followed by sample size. Results highlight the potential of shrinkage-based FC approaches for neuroimaging-based classification