1722 research outputs found
Sort by
LTF Workshop
No full textWe work with community researchers and knowledge keepers from Asiniskaw Īthiniwak communites in northern Manitoaba which are found along the Misinipi, or the Churchill river which is life line connecting RC in the northcommunities primarily with South Indian Lake and Nelson House. Also with university researchers from UW, UM, Lakehead, Alberta, UBC, Ryerson, in many disciplines, oral stories and history, history, archaelogy, picturebook and pb app studies.The Six Seasons Project of Asiniskaw Īthiniwak project is supported in part by funding from the Social Sciences and Humanities Research Council.https://winnspace.uwinnipeg.ca/handle/10680/223
Database Systems Examination and Digital Forensics Tool: The Progress and Limitations
No full textPreprintDatabases play a critical role in many computing systems and applications and provide an excellent source of information for forensics investigations. Also, with many advances in the digital forensics field, several forensic tools can be employed for different purposes in digital investigations. Despite this, many forensics tools have limitations when it pertains to the collection, examination, or analysis of potential evidence from database management systems due to the inherent complexities of handling different database systems. This is particularly true for free or open-source tools that can be used for research purposes, and this limits the development of new tools and solutions for database forensics. To address this and forge a path for the development of new tools for database forensics, this paper highlights some of the available tools that can be used for database forensics, the limitations of using some of these tools, and the challenges of performing database analysis in general. We achieve this through a practical analysis of the tools and examine their capabilities in terms of supporting the forensics analysis of databases found on digital devices. Given their popularity, we consider databases found on both Android and iPhone devices, as well as other data sources. We integrated an analysis of relevant testing reports from the Computer Forensics Tool Testing (CFTT) program to provide a complete picture of the forensic tools. This paper establishes the aspects where these tools can be improved and provides recommendations for handling some of the challenges associated with the forensics analysis of database systems.This work is supported by The University of Winnipeg, Grant ID: 16792.https://ieeexplore.ieee.org/document/1126885
Investigating NoSQL Injection Attacks on MongoDB Web Applications
No full textPreprintModern web applications are increasingly data-intensive and handle a wide variety of semi-structured and unstructured data. Traditional relational databases were not designed to manage such data, and using them often leads to complications in storage, retrieval, and performance degradation. As a result, NoSQL databases have become more popular in recent years. With key advantages such as high performance, scalability, and support for diverse data structures, NoSQL databases are particularly well suited for handling large volumes of data, making them a popular choice for small businesses seeking efficient storage solutions. However, storing sensitive information in these databases makes them targets for attackers aiming to steal data or compromise applications. One of the major security threats to NoSQL databases is NoSQL injection, which attackers exploit to gain unauthorized access. This research investigates how MongoDB-based web applications can be exploited through simulated NoSQL injection attacks, examines their impact, and evaluates mitigation techniques to prevent these security risks. The findings aim to help NoSQL database developers understand the mechanisms of NoSQL injection attacks and implement stronger security measures to protect sensitive data.https://ieeexplore.ieee.org/document/1101196
Tigers on the Prowl: Determining the Distribution and Habitat Requirements of Eastern Tiger Salamander (Ambystoma tigrinum) in Manitoba
No full textManitoba has the only remaining population of Eastern Tiger Salamanders (Ambystoma tigrinum) in Canada. Unfortunately, very little is known regarding its range and habitat in the province. The purpose of this study was to determine key biological and physical habitat requirements that influence the distribution of A. tigrinum in the forested areas of southeastern Manitoba. Fifty-four sites were surveyed in the Sandilands Provincial Forest and surrounding areas using a combination of visual searches and funnel-trapping methods to locate salamander breeding ponds. Much of the Sandilands provides quality habitat for A. tigrinum with little variation between ponds. They are typically sandy, shallow, clear, and most ponds do not contain fish or are unable to sustain a fish population for more than one season. The upland habitat is forested with little agriculture or mining but with some forestry activity. Overall, percent of sand ranged from 85.7% to 99.7%, indicating that the substrate in all ponds is similarly friable. The ponds were mostly small, ranging from 61.38 m2 to 2336.42 m2. All ponds are relatively shallow, ranging in depth from 23 to 300 cm; all sites where salamanders were found at least once were less than 160 cm in depth except for one outlier that is 300 cm. The slope of the bank ranged from -0.06 to -0.80 cm/cm, but salamanders were only found in sites with slopes of less than -0.54 cm/cm. There was a large range in both aquatic vegetation cover (0 to 100% submerged vegetation; 0 to 50% emergent vegetation) and proximity of terrestrial vegetation (minimum distance to forest 0 to 23 m, maximum distance to forest 0 to 312 m; % of shoreline vegetated (0 to 100%), although both ponds with and without salamanders had similar ranges for these vegetation variables (above). Even with the relative homogeneity of the landscape in Sandilands, salamander larvae are not found in all ponds. The results of the analysis identified two key environmental variables: salamanders were more likely to be found in ponds with no fish and with low TDS. Of the 54 total sites, 13 sites had fish present, and of those 13 sites, zero had salamanders present in more than one year. The concentration of TDS in the ponds (the critical value for the split was 207 ppm TDS) also impacted salamander presence. In ponds with high concentrations of TDS (n = 13), approximately nine sites did not have salamanders, while zero sites had salamanders in only one year and four sites had salamanders in more than one year. Ponds with lower TDS concentrations (n = 28) had the opposite pattern of salamander presence; five sites did not have salamanders, four had salamanders in only one year while 19 sites had salamanders in more than one year. With global amphibian decline and the increased risk of climate change, knowledge of local populations of salamanders is important for the conservation of the species, informing policy and recovery strategies.Master of Science in Bioscience, Technology and Public Polic
Researcher Perspectives of Power and Empowerment in Indigenous Community-Based Research
No full textCommunity-based research (CBR), a methodology which aims to shift power dynamics and empower research participants for social justice ends, has gained significant credibility and popularity in recent decades for research involving Indigenous peoples and communities. However, the concepts of power and empowerment are not well-explained in existing CBR literature, with limited description of what power hierarchies in research are, what it means to challenge them, and what it means to empower participants. This is the first study to explore these concepts in-depth through interviews with researchers. As well as contributing a pragmatic overview of many of the understandings and strategies that researchers use in empowerment-focused CBR projects, this research also questions some assumptions underlying researchers’ perspectives to contribute to ongoing critical discussion. As an exploratory case study, rather than defending a particular hypothesis, this research will serve as a foundation for future investigation into power and empowerment in research.This research was supported by the Social Sciences and Humanities Research Council of Canada, Mitacs, and The University of Winnipeg.Master of Arts in Environmental and Social Chang
Assessing the in vitro effects of priority Arctic contaminants using cell-based transcriptomics
No full textPersistent organic pollutants (POPs) are a diverse group of chemicals that resist degradation, bioaccumulate in food webs, and pose risks to human health. Despite international regulation, both legacy and emerging POPs remain widely detected in the environment and human populations. To support the transition toward new approach methodologies (NAMs) that minimize animal use, this study employed human cell–based transcriptomics to investigate the molecular and concentration-dependent effects of POPs. Using human liver (HepG2) and adrenal (H295R) cell models, transcriptional responses to multiple contaminant classes, including organophosphate and organochlorine pesticides, perfluorooctane sulfonic acid, polychlorinated biphenyls (PCB), and polybrominated diphenyl ethers (PBDE), were evaluated. Cytotoxicity assessment, differential gene expression analysis, benchmark dose (BMD) modeling, and biological pathway analysis were integrated to derive transcriptomic points of departure (tPODs) and characterize molecular responses. Concentration-dependent transcriptional changes revealed distinct molecular signatures across chemicals, reflecting diverse modes of action. Benchmark dose modeling of transcriptomic data yielded tPODs that were generally within an order of magnitude of reported apical effect concentrations, demonstrating the relevance of these molecular thresholds for risk assessment. Comparative analysis of PCB-138 between the two cell types indicated cell-specific transcriptional responses, emphasizing the importance of tissue context in evaluating POP toxicity. Overall, this work demonstrates the value of in vitro transcriptomic profiling for mechanistic and quantitative toxicity assessment of POPs. Integrating gene expression and dose–response modeling approaches provides an animal-free framework for defining biologically meaningful toxicity thresholds and advancing next-generation chemical risk assessment.Master in Environmental and Social Chang
Tensions and Transformations: Equity, Diversity, and Inclusion in a Diverse Urban School Division
No full textThis study examines how equity, diversity, and inclusion (EDI) are understood and implemented within a large, diverse, urban school division in Canada. Using a mixed-methods approach and grounded in theoretical perspectives, including critical race theory, intersectionality, the ecology of inclusive education, and collective efficacy theory, we gathered insights from surveys and focus groups with educators, administrators, clinicians, educational assistants, caregivers, and students. Findings suggest that, while support for EDI is broad, the implementation is inconsistent and shaped by systemic barriers, resource constraints, and unclear leadership. We highlight the need for shared language, coordinated supports, and ongoing professional learning and instructional redesign. We offer a model for examining EDI at a systems level and share lessons to guide sustained transformation in other school organizations.This study was funded by the University of Winnipeg through a Major Research Grant. The support provided by this grant is gratefully acknowledged.https://scholarworks.waldenu.edu/jerap/vol15/iss1/30
Prediction of Pea Yield and Nodulation from Proximal Field and Root Imaging
No full textPulse crops such as pea are an important source of plant-based protein as well as being a key component of sustainable agriculture through biological nitrogen fixation. However, phenotyping methods for predicting yield are lacking and traditional root nodule counting is slow and prone to human bias. This thesis explores the application of deep learning to two critical phenotyping challenges in pulse crops: temporal yield classification for breeding program efficiency and root nodule detection for nitrogen fixation capacity assessment. While both tasks leverage computer vision to extract meaningful information, they differ fundamentally in scope and complexity - nodule detection addresses a controlled object detection problem, whereas yield classification faces dynamic, environment-dependent variation across growing environments. For yield classification, we addressed the prediction of high and low yield outcomes using sequential images captured throughout two growing seasons (2023 and 2024). The findings showed that environment conditions producing clear phenological differentiation are more important than the dataset size, achieving 0.89 F1-score on 2024 data despite having less training data than 2023 (0.67 F1 score). Multi-year training under-performed as compared to single year with F1-score collapsing from 0.67 to 0.08 on 2023 test data. Explainable AI using attention weights and integrated gradients showed that the models acquired temporal patterns that were year-specific and not generalizable features, which means that conflicting phenological signals and differences between environments can lead to negative transfer. Also, explainable AI highlighted mid-season to late season as a critical window for yield prediction. DINOv2 self-supervised features consistently outperformed supervised ResNet-152 features with a balanced contribution from RGB and vegetative index - green leaf index. For root nodule detection, we trained and tested YOLOv12 and Mask R-CNN architectures on pea roots which were grown in controlled-environment rhizoboxes. YOLOv12 significantly outperformed in all measures: precision (0.595 vs. 0.46), recall (0.71 vs. 0.60), dice coefficient (0.65 vs. 0.52) and equal error rate (0.36 vs. 0.48). Implementation of explainable AI like Grad-CAM++ revealed that YOLOv12 focused on nodule patterns, providing transparency for model deployment. Overall, this thesis demonstrated the promise of deep learning in agricultural phenotyping. The findings provide a foundation for developing scalable, trustworthy AI systems that accelerate crop improvement while enhancing the interpretability of agricultural data.Mitacs; National Research Council of Canada, SaskatoonMaster of Applied Computer Science and Societ
Hybrid Fingerprint Classification using Deep Learning and Sobel Feature Fusion
No full textPreprintFingerprint classification serves as a critical preprocessing step in biometric and forensic identification systems. By categorizing fingerprints into distinct types, such as Thumb, Index, Middle, Ring, and Little, the process significantly reduces the search space, thereby enhancing the efficiency, s peed, and accuracy of matching algorithms. This is particularly valuable in large-scale identification tasks and law enforcement applications. However, achieving accurate classification remains challenging due to high visual similarity between classes, intra-class variability, and imbalanced datasets. In this study, we present a hybrid fingerprint classification framework, which combines deep learning-based embeddings with handcrafted Sobel edge features to improve both model robustness and interpretability. Three state-of-the-art architectures—EfficientNet: B0, ResNet50, and Vision Transformer (ViT-B/16), were employed to extract semantic feature representations, which were fused with Sobelbased statistical edge descriptors. The models were trained and evaluated on the SOCOFing dataset using a stratified 70-15-15 split, with class imbalance addressed via oversampling using WeightedRandomSampler. EfficientNet-B0 achieved the best performance, with a test accuracy of 99.19%, F1-score of 99.18%, and recall of 99.19%. The models’ reliability and precision across fingerprint c lasses were further validated through confusion matrix analysis and visual prediction results, demonstrating the effectiveness of the proposed hybrid approach for fine-grained fingerprint classification.This work is supported by The University of Winnipeg, Grant ID: 16792.https://ieeexplore.ieee.org/document/1129979
Summer Roosting and Foraging Habitat of Endangered Little Brown Bats (Myotis lucifugus) in Central Canada
No full textLittle brown bats (Myotis lucifugus) are listed as endangered in Canada and by the International Union for the Conservation of Nature (IUCN) due to population declines caused by the fungal disease white-nose syndrome (WNS). Under the Species at Risk Act, conservation efforts must focus on identifying and protecting critical habitats essential for the survival and recovery of listed species. While current efforts for little brown bats emphasize hibernation sites, designating maternity roosts as critical habitats are equally important. Despite some existing data, maternity roosts are not yet designated as critical habitats, but it is crucial as maternity roosts support bats’ recovery from WNS in the spring and successful reproduction and juvenile development in the summer. Understanding both roosting and foraging habitat selection, as well as drivers of foraging behaviour, is important for developing recovery plans, as these habitats are often spatially associated (Balzer et al. 2023). The broad objectives of my thesis are to characterize summer maternity roosts, better quantify roost switching and foraging habitat selection, and better quantify environmental variables that influence bats’ nightly activities. To achieve this, I tagged 30 lactating little brown bats in the summers of 2021 and 2022 (n=15 bats/year) in Ontario, Canada, and tracked them to roosts to identify their roosting habitat selection and quantify roost-switching behaviours. I used data from 2021 (n=15 bats) to understand foraging habitat selection and the influence of factors such as minimum nightly temperatures, maximum wind speed, and air quality on nightly activities and home-range sizes. My Chapter 2 results show that bats predominantly preferred buildings and bat houses over trees, with the largest groups found in buildings, while bats in trees were always solitary. Regardless of roost types, bats preferred structures close to water with southern exposure, possibly to reduce commuting costs to food and drinking areas and to maximize heat gain. These findings should help identify geophysical attributes important for the designation of maternity roosts as critical habitats. Bats in buildings switched roosts significantly less than when they roosted in bat houses or trees, but still switched more than expected, suggesting that management strategies should focus on networks of suitable roosts, rather than individual structures, to meet bats’ roosting requirements. My Chapter 3 results show that bats preferred foraging over wetlands, open water, anthropogenic areas, and forest edges, with bats having larger home-range sizes than observed in comparable studies of this species. Taking advantage of variation in air quality due to wildfire near my study area, I found that bat nightly activities and home-range sizes were significantly affected by air quality, with bats being less active and having smaller home ranges on nights with poor air quality. I found no effect of minimum nightly temperatures or maximum wind speeds on foraging behaviour. These findings underscore the importance of protecting both roosting and foraging habitats to support the recovery of endangered little brown bats.Natural Sciences and Engineering Research Council (Canada); Environment and Climate Change CanadaMaster of Science in Bioscience, Technology, and Public Polic