21793 research outputs found
Sort by
Stability of oxylipins stored on biocompatible solid-phase microextraction (SPME) devices
Get PDFABSTRACT
Stability of oxylipins stored on biocompatible solid-phase microextraction (SPME) devices
Oluwatosin Kuteyi, PhD.
Concordia University, 2025
Oxylipins are lipid mediators involved in inflammation, immunity, and oxidative stress. Accurately measuring oxylipins in biospecimens is analytically challenging due to their poor stability and susceptibility to enzymatic and non-enzymatic reactions during sampling, storage, and transportation. Recently, in vivo solid-phase microextraction (SPME) has been introduced as an effective method for direct sampling and extraction of oxylipins from biological tissues and fluids. Although in vivo SPME protects oxylipins from enzymatic degradation, it is not known how the storage in the SPME coating affects oxylipin stability and susceptibility to non-enzymatic reactions such as autoxidation, hydrolysis, and isomerization. The objective of this thesis was to evaluate the stability of oxylipins on SPME devices post-extraction and investigate whether the use of antioxidants, such as butylated hydroxytoluene (BHT), is useful in minimizing degradation processes. To evaluate the effect of 3-freeze-and-thaw cycles (3-FT), and 18-day room temperature (RT) storage on the stability of oxylipins on SPME devices, oxylipins were extracted from standard solutions or citrated human plasma samples using hydrophobic lipophilic balance (HLB) SPME devices and analyzed by C18 liquid chromatography-high-resolution mass spectrometry (LC-HRMS). The pre- and post-extraction loading methods for antioxidants were successfully developed in order to investigate the ability of BHT to minimize oxylipin autooxidation during storage of SPME devices. Finally, degradation products of selected unstable oxylipins were comprehensively mapped by forced degradation studies including photooxidation (365 nm for 5 and 7 days), copper sulphate oxidation, and elevated temperatures (37°C and 50°C for 3 days). In conclusion, this is the first study to characterize the stability of oxylipins on SPME devices, demonstrating how SPME can effectively improve stability during sample storage, handling, and shipping even without the use of BHT. Importantly, these results also show how the degradation of unstable oxylipins can impact the accurate measurement of stable oxylipins and provide novel insight into major degradation products of PUFAs and selected unstable oxylipins
Strategic Resource Planning in Gold Mining: Optimizing Supply Chain Management with Neural Networks-Based Gold Price Forecasting
No full textStrategic resource planning is crucial for optimizing supply chain management and ensuring efficient operations. This study aims to enhance strategic planning in gold mines by leveraging advanced gold price forecasting models. By predicting future gold prices accurately, mining companies can better plan their extraction, processing, and distribution activities, thereby improving overall supply chain efficiency. We employed various advanced forecasting models, including Unidirectional and Bidirectional Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Artificial Neural Network (ANN), to predict gold prices and analysed how these predictions can inform strategic decisions in the gold mining supply chain. Our approach includes evaluating the performance of these models using metrics such as root mean squared error (RMSE), mean absolute percentage error (MAPE), and mean absolute deviation (MAD). Results show that Artificial Neural Network (ANN) performed best, with the lowest (0.3514), RMSE (0.5928), and MAPE (0.34%), while Bidirectional Gated Recurrent Unit (GRU) was the poorest performer with an of 88.5474 and MAPE of 6.94%. The feature selection process, facilitated by Recursive Feature Elimination (RFE), identified critical predictors such as 'High,' 'Low,' 'Volume,' and various external market factors. Optimizing model parameters through techniques like grid search and cross-validation further improved model accuracy. Additionally, advanced forecasting models, particularly Artificial Neural Network (ANN) and Convolutional Neural Network (CNN), proved highly effective in refining gold mining companies' resource planning and supply chain management strategies, providing critical managerial implications for navigating the dynamic and volatile gold market
Towards LLM-Driven Code Generation: The Impact of Process Models and Non-Functional Requirements on Software Development
Get PDFResearch on LLM-based code generation is growing but often overlooks the impact of Software Engineering (SE) knowledge, such as software process models and non-functional requirements (NFRs). This thesis explores how integrating SE practices with LLMs can help bridge this gap.
We first propose FlowGen to explore the impact of software process models on code generation. We assign LLM agents to different development roles and study three models: FlowGenWaterfall, FlowGenTDD, and FlowGenScrum. We evaluate FlowGen using GPT-3.5, comparing it against several baselines on four code benchmarks (i.e., HumanEval, HumanEval-ET, MBPP, and MBPP-ET). Our results show that FlowGenScrum outperforms the other process models, achieving a 15% improve- ment in Pass@1 over RawGPT on average. Integrating a state-of-the-art technique (i.e., CodeT) further boosts Pass@1 scores. Our findings also show that development activities impact code qual- ity differently, and FlowGen enhances result stability across LLM versions and temperature settings.
Secondly, we investigate how variations in developer behavior affect how LLMs address NFRs (e.g., expressing the same NFR using different wording). Robust LLMs should generate consistent code despite such variations. We propose RoboNFR to evaluate LLM robustness in NFR-aware code generation across four key dimensions—code design, performance, readability, and reliabil- ity—using three methodologies: prompt variation, regression testing, and diverse workflows. Our experiments show that RoboNFR effectively reveals robustness issues in tested LLMs. Overall, across the three methodologies, incorporating NFRs tends to reduce Pass@1 scores while improv- ing NFR-specific metrics, but also increases the standard deviation in both correctness and quality.
This thesis highlights the significant influence of software process models and NFRs, empha- sizing the need for future work to incorporate such SE knowledge in the era of LLMs
Practicality of Sub-Linear Algorithms for Distributed Shortest Paths
Get PDFThe computation of Single-Source Shortest Paths (SSSP) is a fundamental problem in distributed
networks. This thesis explores efficient SSSP solutions within the synchronous CONGEST
model, where communication occurs in rounds in which each node can send O(log n) information to
each of its neighbors, where n is the size of the network. We focus on Elkin’s algorithm, notable for
being the first algorithm to achieve sub-linear round complexity for the problem. This work makes
several contributions: we clarify many details in the implementation of Elkin’s algorithm, enhance
its performance using pipelining techniques, and introduce two novel variants, Elkin-R (probabilistic
virtual node selection) and Elkin-D (distance-based selection). Through extensive simulations on
various network topologies, we compare these algorithms against Bellman-Ford based on the number
of rounds, message count, and clock time. Our findings reveal that specific configurations of
the Elkin variants consistently reduce message overhead compared to Bellman-Ford. Importantly,
for larger or denser graphs, these variants can also surpass Bellman-Ford in round complexity and
overall computation time, demonstrating their potential for practical distributed SSSP computation
Being a Dawoodi Bohra Woman: Community, Religious Agency, and Social Media
Get PDFThis thesis considers the everyday lives of pious Dawoodi Bohra women to understand how these women construct their subjectivity and embed their agency within a religious community to constitute a self that is multilayered and challenges the way religious women’s lives have been understood in scholarship. The thesis analyses the women’s lived experience and understandings of embodiment, gender roles, mothering, and other everyday activities. The role of self-representation on social media in constructing subjectivity is also considered. This thesis argues that the lives of women in conservative religious contexts are not necessarily “encapsulated by narratives of subversion” (Mahmood 2005, 9) and that agency needs to be examined independently of such ideas and and in its specific social and cultural context. The thesis shows how religious women deploy ‘creative conformity’ (Bucar 2011) to negotiate religious and community norms in their everyday lives while remaining within the boundaries of the Dawoodi Bohra community
Data-Driven Security Monitoring for False Data Injection Attacks on Subsynchronous Damping Controllers in PMSG-based Wind Farms
Get PDFThe integration of Permanent Magnet Synchronous Generator (PMSG) Wind Turbines (WTs)
into the power grid requires advanced control strategies to maintain stability and damp oscillations,
particularly under weak grid conditions. These strategies, along with wind farm control loops, often
rely on data transfer and information provided by communication networks.
However, the cyber layers used for such data transfer make the entire network prone to various
cyber threats, such as False Data Injection Attacks (FDIAs). These attacks can compromise power
grids’ stability and operational integrity and result in blackouts. On this basis, we highlight the
vulnerability of communication links of PMSG-based WF to FDIAs and propose a data-driven
detection system developed based on a Convolutional Neural Network (CNN) to identify threats.
First, FDIAs are introduced in the simulation by manipulating the communicated signals between
the SCADA system of the wind farm and WT controllers. Second, the CNN model is trained
using grid and wind farm data in various operating conditions to detect FDIAs, distinguishing them
from normal operational variations. To evaluate the performance of the proposed detection method,
it is tested in a wind farm connected to a power system and compared with traditional data-driven
detection methods based on K-Nearest Neighbors (KNN), Random Forest (RF), and Extreme Gradient
Boosting (XGBoost). The results demonstrate that CNN achieves high detection rates with
minimal false positives, validating its efficiency in detecting FDIAs in grid-connected wind farms
Stability and Cooperation in International Environmental Agreements: Essays on Coalition Structures, Ethical Incentives and Leadership
Get PDFThis thesis addresses the multifaceted challenges of global climate cooperation through a comprehensive analysis of international environmental agreements. It examines three critical dimensions: the formation and stability of multiple international environmental agreements, the impact of ethical considerations on climate governance, and the role of leadership in fostering cooperation on net emissions. These themes provide fresh insights into the stability and effectiveness of environmental agreements.
The first essay explores the existence and stability of multiple international environmental agreements within a two-stage non-cooperative coalition formation game framework. Utilizing a partition function form, the model enables the formation of multiple coalitions. This approach captures inter-coalition externalities and allows for a refined analysis of coalition dynamics. Findings reveal a unique stable coalition structure characterized by multiple coalitions, achieving greater pollution reduction and welfare gains compared to a single coalition outcome.
The second essay investigates the role of ethical incentives in emissions reduction and the stability of climate agreements by incorporating into a classic model a cost-function that reflects the evaluation of a country's deviation from the global emissions norm to analyze the effects on emissions, cooperation, and welfare in simultaneous coalition formation and leadership game settings. The results indicate that uniform ethical motivations across all countries are more effective at curtailing emissions and improving welfare than partial commitments. Such broad ethical alignment within a leadership framework yields higher level of cooperation. The findings underscore the critical importance of promoting ethical alignment among all countries in fostering stricter environmental policies.
The third essay examines the role of leadership in fostering cooperation on net emissions through a non-cooperative leadership game, in which decisions on net emissions and abatement are strategically decoupled into two distinct stages. It explores how variations in abatement technology and environmental damage costs affect both the size of a stable coalition and the gains from cooperation. The findings suggest that the coalition's stability is robust to changes in these key parameters. Moreover, the results indicate that when abatement is costless, countries invest substantially in reducing environmental pollution while retaining economic benefits. In contrast, when abatement becomes costly, countries are incentivized to reduce emissions directly, which lowers total welfare gains.
Together, these essays contribute to the literature on international environmental agreements by offering novel perspectives on coalition formation, leadership dynamics, and the integration of ethical incentives in addressing climate change and promoting global cooperation on pollution reduction
Spatio-Temporal Urban Analytics: Data-Driven Building Occupancy Estimation and Graph Deep Learning for Mobility Demand Forecasting
No full textUnderstanding urban-scale building occupancy and commuting flow patterns is critical for sustainable city planning, energy efficiency, and transportation optimization. However, existing Urban Building Energy Modeling (UBEM) frameworks rely on standardized occupancy schedules that fail to capture spatial and temporal variations, leading to significant inaccuracies in energy consumption estimates.
This thesis introduces a data-driven approach to improve urban-scale occupancy estimation and commuting flow prediction by integrating mobility data with machine learning techniques. A Transportation-Informed Building Occupancy (TIBO) framework is developed to estimate dynamic building occupancy profiles using large-scale transportation datasets, including metro, bus, bike-sharing, and traffic flow data. TIBO-based occupancy profiles significantly enhance UBEM simulation accuracy compared to conventional ASHRAE schedules, reducing errors in energy demand predictions.
Since transportation-based occupancy estimation frameworks depend on transportation data as input, enhancing transportation demand and flow prediction improves building occupancy prediction. Therefore, this thesis presents a multi-task spatiotemporal deep learning framework for short-term bike-sharing demand prediction at the station level. By incorporating historical demand patterns, meteorological data, and a dynamically evolving semantic adjacency graph, the model jointly predicts bike pick-ups and drop-offs, addressing system imbalances.
Additionally, a novel geographic-semantic graph-based model for commuting flow prediction is introduced. By leveraging Graph Convolutional Networks and Graph Attention Networks, the model captures both geographic adjacency and semantic connectivity, providing precise estimation of urban mobility patterns. The proposed approach outperforms existing models, improving accuracy in forecasting commuting demand across city zones.
This research provides a holistic methodology for modeling urban dynamics, contributing to more accurate energy simulations, better-informed transportation planning, and smarter, sustainable cities
Depth and Segmentation Aware frameworks for Multiple Object Tracking
Get PDFMulti-Object Tracking (MOT) remains a challenging problem, particularly in crowded scenes with occlusion, appearance ambiguity, and non-linear motion. Conventional MOT frameworks often rely on appearance-based Re-Identification (Re-ID) and Intersection-over-Union (IoU) of object bounding boxes for object association. However, these cues become unreliable when objects are visually similar or overlapping, and computing pixel-level IoU for segmentation masks can be computationally expensive.
In this thesis, we propose two complementary MOT frameworks that incorporate monocular depth and segmentation cues to improve robustness in association. The first zero-shot depth-aware framework is training-free and introduces a Hierarchical Alignment Score (HAS), a novel metric that combines coarse bounding box IoU with fine-grained mask-level IoU using promptable segmentation. This hierarchical formulation improves matching precision in cluttered or occluded scenes.
The second framework avoids computing segmentation IoU altogether. Instead, it leverages a self-supervised encoder to fuse and refine depth-segmentation features into temporally stable embeddings, which are then used as an additional similarity signal in the association process. This reduces computational overhead while improving robustness to noise and appearance variation.
Both approaches operate under the efficient Tracking-by-Detection (TBD) paradigm and extend conventional 2D association strategies with spatially expressive cues. Evaluations on DanceTrack and SportsMOT benchmarks with non-linear motion demonstrate competitive performance, highlighting the utility of depth and segmentation as underutilized, yet powerful, cues for robust non-linear MOT
High-Resolution Pollen and Charcoal Records from Fish Lake, New Brunswick, Canada
No full textThe Acadian forest is a mixed-wood forest covering the Canadian provinces of New Brunswick and Nova Scotia. Picea rubens (red spruce) is its signature species which has been prominent for circa 2,000 years. To our knowledge, no high-resolution pollen analysis has been done in New Brunswick, and no lacustrine charcoal analysis. We extracted a 124 cm surface core from Fish Lake, (46° 8’ 38.32’’N, 66° 53’ 12.64’’ W), near Fredericton, New Brunswick, Canada. A BACON age-depth model, based upon 4 14C and 15 210Pb dates, showed that the bottom of the core dated to AD 890. We performed a high-resolution pollen analysis on this core at a ~10 year resolution, with 125 samples in all. Over the last millennium, there were declines in Betula (birch), Tsuga canadensis (eastern hemlock) and Fagus grandifolia (American beech), together with increases in Pinus sp. (pine), Abies balsamea (fir) and Picea sp. (spruce), including P. rubens. CONISS showed that the Medieval Climate Anomaly (MCA – AD 900-1400), Little Ice Age (LIA – AD 1400 –1850), and the European settlement period were clearly demarcated in the pollen record. A rise in Ambrosia (ragweed) marked early Acadian French agriculture at ~AD 1680. The last 300 years of the European period showed increases in Poaceae (grasses), Ambrosia, and other herbs, and declines in Pinus sec. haploxylon (eastern white pine). Charcoal analysis showed that natural forest fires had a continuous presence over the past millennium. The paleo-fire record showed higher fire frequency during the MCA than in the LIA. WA-PLS was used to reconstruct spring temperatures. The reconstruction showed that the MWP had an average spring temperature of 3.2 °C, and the LIA had an average spring temperature of 2.2 °C