University of British Columbia
University of British Columbia: cIRcle - UBC's Information RepositoryNot a member yet
88081 research outputs found
Sort by
Efficient dynamic analysis of Android applications
No full textDespite efforts to identify low-quality applications (apps) such as apps that crash or leak sensitive user information, the official Android Google Play app store continues to be infected by such apps. This causes issues ranging from negative user experience to millions of dollars in financial losses.
One approach that the store uses to check apps is dynamic program analysis, in which an app is executed, and its runtime behavior is observed. However, existing dynamic analysis techniques often sacrifice the comprehensiveness of the analysis (i.e., do not report all interesting behaviors) in favor of efficiency. This is necessary since Android devices have limited resources, and the system aborts unresponsive or memory-hungry apps, terminating slow analysis prematurely.
In this thesis, we propose techniques to improve the comprehensiveness of dynamic analyses while keeping them efficient. Specifically, we focus on two of the most prominent (yet heavy) analysis techniques: (a) slicing, which retrieves a slice: a set of statements that affect the execution of a specific statement (e.g., a buggy line of code), and (b) taint analysis, which identifies whether specific data flows out.
Our proposed slicing approach provides more comprehensive slices (more relevant statements) than the state-of-the-art while being ten times more efficient. Our taint analysis approach can report full information flow paths, unlike state-of-the-art approaches that only report flow endpoints. We also showcase the usefulness of reporting full paths by introducing a new approach that classifies suspicious and legitimate information flow paths with high accuracy while remaining efficient.
We also introduce new datasets to aid with the evaluation of our approaches for real use-cases and in realistic settings. For slicing, we introduce the first dataset of manually generated slices for real bugs from real apps. For taint analysis, we create a dataset of apps with malicious code that entered the Google Play Store. We characterize their attacks and propose a novel representation for summarizing their activation methods. We use information flows from this dataset in evaluating our path classification approach.
Finally, we implement our approaches in open-source tools and make them and our datasets available to the research community.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat
Rapid genomic monitoring of municipal wastewater and treatment facilities with Nanopore sequencing
No full textThe full abstract for this thesis is available in the body of the thesis, and will be available when the embargo expires.Applied Science, Faculty ofCivil Engineering, Department ofGraduat
Same deck of cards, different shuffle : conceptualizing financial literacy in the B.C. K-12 curriculum, 1988-2018
No full textConceptualizations of financial literacy are ubiquitous and have received considerable attention in the literature (Arthur, 2012a; Bay et al., 2014; Pinto & Coulson, 2011) and mainstream media (Hughes, 2025; Steffenhagen, 2011). Attempts to define these concepts reveal the absence of one singular moment or event with the universal authority to validate one as be(com)ing financially literate. Establishing meaning(s) about financial literacy is contextual yet they are often falsely presented as value neutral. As a parent and chartered professional accountant teaching in higher education, I was curious to know how the British Columbia Ministry of Education conceptualized financial literacy in its K-12 education policies during two education reform initiatives spanning 1988 to 2018.
Using Carol Bacchi’s (2016) ‘What’s the Problem Represented to Be?’ methodology opened space(s) to critically analyze how the British Columbia Ministry of Education constituted the financial literacy problem within its education policy discourses and curriculum as a policy instrument. Interrogating the Ministerial interpretations and/or representations within the policy texts employed to frame the financial literacy problem facilitated the unravelling of assumptions that undergirded this problem definition and its proposed solutions.
This study found that the conceptualization of financial literacy intersected within three curricular learning areas, namely, applied skills, career and personal planning, and mathematics where courses containing financial literacy were compulsory for graduation and prioritized by way of a royal proclamation affirming its significant importance in K-12 education. Influenced by the OECD sponsored Program for International Student Assessment, students’ performance in financial literacy was measured to identify remedial approaches that addressed the perceived deficit thinking assumptions that students lacked financial (cap)abilities, qualities, and/or skills. Recurring notions of financial literacy centred on homogeneous, skills-based, and standardized applications that brought into question the social-cultural silences embedded in the curriculum. Without consideration of the multidimensional constructs of financial literacy, ones shaped by an individual’s ideological background/s, identity/ies, and belief/s, the curriculum diminishes the pluralistic voices of young learners. Gaining a deeper understanding of the varying conceptualizations of financial literacy and how they may conflict and/or contradict with those in the curriculum, brings forth opportunities to enrich student learning about financial literacy.Education, Faculty ofEducational Studies (EDST), Department ofGraduat
Software solutions for mitigating physical attacks against robotic autonomous vehicles
No full textRobotic Autonomous Vehicles (RAVs) rely on onboard sensors for perception and autonomous decision making. However, RAVs are vulnerable to physical attacks, such as GPS spoofing and gyroscope manipulation, that exploit physical channels to compromise sensors. These attacks pose serious safety risks and can lead to widespread damage and mission failures. Traditional software security methods, such as cryptography or memory isolation, are ineffective against physical attacks.
This thesis presents five comprehensive software solutions to secure RAVs against physical attacks. These solutions do not require any hardware modifications, and can be seamlessly integrated into RAV's existing software stacks. (1) RAVAGE introduces a configurable software tool that emulates physical attacks in a realistic and reproducible manner. It provides a software setup for benchmarking the impact of physical attacks and validating defense techniques.
Building on this foundation, this thesis presents attack detection, diagnosis, and recovery techniques, which enable RAVs to operate safely even under adversarial interference.
(2) PID-Piper introduces a Feed-Forward controller (FFC) that runs in tandem with the RAV’s primary controller and monitors it to detect attacks. The FFC uses a system model instead of corrupted sensor inputs; thus, it remains robust to attacks. Upon attack detection, the FFC takes over control to recover the RAV back to its desired state and enables mission completion. (3) DeLorean performs diagnosis to isolate the compromised sensors, and uses historical data to derive safe recovery actions. Its diagnosis-guided recovery mechanism enables reliable operation even under multi-sensor attacks. (4) SpecGuard combines multi-objective policy learning with adversarial training to achieve resilience against sensor perturbations. SpecGuard not only ensures mission completion under attack but also enforces safety specifications, such as avoiding collisions or preventing entry into restricted zones, during recovery. (5) ARMOR introduces a two-staged reinforcement learning framework that enables learning attack-resilient control policies. It first leverages attack-aware privileged information to accelerate robust policy learning in simulation, and then uses transfer learning to adapt the policy for real-world deployment.
By developing comprehensive and practical software solutions for physical attack detection, diagnosis, and recovery, this work enables the deployment of RAVs in a wide range of safety critical applications.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat
Proslavery representations of enslaved people and their management in the literatures of ancient Greece, Rome, and the 17th-19th-century Atlantic : a comparative perspective
No full textThis dissertation compares how proslavery authors in ancient Greece, Rome, and the 17th-19th-century Americas and Caribbean depicted enslaved characters and their management. The comparison shows that ancients and moderns handled and thought about enslaved people in dramatically different ways. The fiction of proslavery Greeks and Romans represented enslaved and free people as equally human, possessing the same desires for material goods, autonomy and authority at work, and freedom. Ancient slaveholders portrayed enslaved characters in this way to justify a comparatively open system of slavery, one in which they were free to manage the enslaved by offering the objects of those desires as incentives for loyalty and diligence. Ancient slaveholders embraced a humanist ideology to prolong this open social system. Proslavery fiction in the modern Atlantic represented the exclusively black enslaved population as inferiorly human, as needing and wanting enslavement to white men, because their health and happiness were contingent on it. English and Portuguese authors portrayed enslaved characters in this way to justify more closed social systems, ones which tightly constrained black people’s access to material privileges, civil rights, and freedom itself. Modern slaveholders adopted an ideology of white supremacy to prop up a racial slavery. I argue that this difference between ancient and modern societal structures and ideologies rules out applying racism to Greek and Roman slavery.
The first chapter of the dissertation outlines the comparative historical methodology with which I study slavery and racism. I adopt the framework put forth by black radical historians and critical race theorists who follow the model of W. E. B. Du Bois. I trace the history of that framework from David Walker in the early 19th-century to the present, and I explain why that framework has tended to be passed over in recent Classics scholarship. The following chapters use this black radical framework to perform close readings of ancient and modern proslavery fiction. I demonstrate that their authors resolve plots involving enslaved characters such that the routines propagandize in favour of the specific idiosyncrasies of the authors’ respective slave systems.Arts, Faculty ofAncient Mediterranean and Near Eastern Studies, Department ofGraduat
Making movement matter : an implementation dose-response evaluation of a physical literacy program for children
No full textBackground: Physical literacy (PL) is defined as the motivation, confidence, physical competence, knowledge, and understanding to value and take responsibility for engagement in physical activities for life. Given concerns with child and youth physical inactivity, the concept of PL is gaining interest as a foundation for greater PA participation. A school-based PL program called Make Movement Matter (MMM) was implemented in one elementary school district. The objective of this study was to assess PL changes over time and assess if there was a relationship between implementation dose and changes in PL. A second objective was to assess gender differences in PL changes over time. Methodology: The program was evaluated using a longitudinal dose-response cohort study design, as the PL of a cohort of grade two students in 14 schools was assessed at three time points over two years. PL was measured at each time point using two Physical Literacy Assessment for Youth (PLAY) tools: PLAYfun (measure of children’s fundamental movement skills) and PLAYself (measure of children’s perceptions of PL and attitudes towards PA). Implementation dose was tracked through school checklists completed after each visit by PL mentors. The study was approved by the University of British Columbia Behavioural Research Ethics Board (H21-03586). Results & Conclusion: PL (PLAYfun and PLAYself) was found to significantly change over time (p > .001), with implementation dose having a significant effect for PLAYfun (p < 0.001) and no effect for PLAYself (p = 0.526). Gender was tested in both models, and no significant differences were found (PLAYfun: p = 0.102; PLAYself: p = 0.197). This is the first study to examine the relationship between implementation dose and PL outcomes. As higher dose of implementation was associated with greater changes in PLAYfun, positive changes in PL may be partly attributed to the MMM program. These findings may have implications for future practice in implementing school-based PL interventions.Education, Faculty ofKinesiology, School ofGraduat
Prior knowledge-driven model pre-training and fine-tuning for ophthalmic image analysis
No full textDeep learning has become a cornerstone in ophthalmic image analysis, enabling automated disease detection, grading, and segmentation across modalities such as fundus photography, OCT angiography (OCTA), and fundus fluorescein angiography (FFA). However, its deployment faces major challenges: limited annotated datasets, the need for high-resolution inputs to capture subtle lesions, and the inefficiency of conventional fine-tuning strategies for large pre-trained models (LPMs). This thesis addresses these challenges by developing and evaluating novel pre-training and fine-tuning frameworks tailored for ophthalmic imaging.
We first establish a strong and reproducible baseline for ophthalmic classification and segmentation using Vision Transformers (ViTs). A systematic grid search identifies a robust training pipeline that generalizes across diverse datasets, highlighting the sensitivity of performance to choices such as input resolution, sampling, and optimization strategies.
Building upon this baseline, we introduce SSiT (Saliency-guided Self-supervised Image Transformer), a prior knowledge-driven pre-training framework that integrates saliency maps into contrastive learning. By guiding models to focus on diagnostically relevant retinal regions, SSiT learns transferable and clinically meaningful representations. Experiments on multiple fundus, OCTA, and FFA datasets demonstrate that while conventional self-supervised methods underperform compared to ImageNet pre-training, SSiT consistently surpasses it, confirming the value of embedding clinical priors into self-supervised learning.
To address the prohibitive memory cost of fine-tuning LPMs on high-resolution medical images, we propose FPT+ (Fine-grained Prompt Tuning Plus), a parameter- and memory-efficient fine-tuning method. FPT+ combines asymmetric input design, fine-grained prompts with fusion modules, important token selection, and feature preloading to achieve efficient adaptation. Across fundus and OCTA datasets, FPT+ outperforms state-of-the-art parameter-efficient tuning methods, reducing Graphics Processing Unit (GPU) memory consumption by over 95% while maintaining or improving classification accuracy.
The contributions of this thesis are threefold: (1) establishing a standardized baseline pipeline for ophthalmic image analysis with ViTs; (2) introducing SSiT, which demonstrates the importance of saliency-driven pre-training for learning clinically relevant representations; and (3) proposing FPT+, a memory- and parameter-efficient fine-tuning strategy that enables high-resolution transfer learning on resource-limited hardware. Together, these advances provide a unified framework for efficient and generalizable ophthalmic AI, with potential applications in clinical decision support and large-scale screening.Applied Science, Faculty ofBiomedical Engineering, School ofGraduat
Ice nucleation induced by organic matter at hydrophobic interfaces
No full textWater droplets may remain in a liquid metastable state when cooled to temperatures below
0 °C. Organic materials, such as proteins, non-proteinaceous biopolymers, or plastics, can
initiate the freezing of supercooled droplets at various temperatures, ranging from –2 °C to
–25 °C. Hence, these organic materials can impact the phase state of atmospheric clouds, the
formation of ice crystals in biological organisms, and the ice formation in the aviation and
cryo-preservation industry. The mechanisms of ice nucleation induced by organic materials
remain poorly described, particularly regarding the specific sites where nucleation originates
and the role of hydrophobic interfaces, such as the air-water interface or the plastic-water
interface. These uncertainties limit our ability to predict the ice nucleation temperatures of
organic matter in aerosol-cloud interactions, biological organisms, and technological applications.
This thesis addresses this knowledge gap by (i) applying high-speed cryo-microscopic
imaging to identify the onset locations of freezing in aqueous droplets, (ii) characterizing
the size of ice nucleating nanoparticles with a microscopic scattering technique, and (iii) analyzing
surface properties of materials (topography and hydrophobicity) with atomic force
microscopy and contact angle measurements. The results showed that bacterial proteins triggered
freezing at hydrophobic interfaces, such as the air-water interface of the droplet or
the bacterial membrane. In contrast, non-proteinaceous biopolymers formed nanoparticles,
which triggered freezing immersed in aqueous droplets. The dominant ice nucleation onset
location for plastic materials was at the plastic-water-air contact line. Based on the mechanistic
understanding, surface modification techniques were applied to influence the surface
hydrophobicity and, therefore, the freezing temperatures of plastic interfaces used for ice nucleation experiments. In addition, ice nucleating materials were found to be an innovative
tool in the field of point-of-care diagnostics to detect biomarkers using the freezing of water
as an easily detectable signal. Overall, this thesis adds fundamental knowledge of the role
of hydrophobic interfaces, such as the air-water interface of droplets, the nanoparticle-water
interface, or the plastic-water interface to our understanding of heterogeneous ice nucleation
induced by organic materials and extends the list of potential applications of ice nucleating
particles with point-of-care diagnostics.Science, Faculty ofChemistry, Department ofGraduat
Supervised additive models can decompose and extrapolate single-cell expression aligned with known gene programs
No full textThe full abstract for this thesis is available in the body of the thesis, and will be available when the embargo expires.Science, Faculty ofGraduat
Defining the gene regulatory network of bone morphogenetic protein signaling in chondrogenesis
No full textChondrogenesis is a developmental process that is central to the formation of cartilage and most bone, during which an initial cartilaginous template gradually ossifies in a process termed endochondral ossification. Bone morphogenetic protein (BMP) signaling is a highly conserved signaling pathway that induces a gene regulatory network (GRN) which drives chondrogenesis and endochondral ossification, with disruption of BMP signaling leading to severe skeletal defects. Despite its importance, it remains unclear the specific molecular mechanisms by which BMP signaling exerts its actions during chondrogenesis. This thesis seeks to define the BMP-activated GRN during chondrogenesis, identify targets of BMP signaling during this process, and determine how BMP signaling interacts with other critical regulatory inputs during chondrogenesis. To address these questions, we used murine primary limb mesenchymal (PLM) cells as a model of BMP-activated chondrogenesis. We assessed the transcriptome of BMP4-treated PLMs using RNA-seq, and determined the binding landscape of phospho-SMADs transcription factors (TFs) and the chondrocyte lineage determinant TF SOX9 using the novel sequencing technique CUT&Tag. We found that BMP signaling is required to activate a complete and robust chondrogenic program. SMAD target genes are predominantly comprised of genes with pleiotropic and indirect functions in chondrogenesis, such as cell adhesion. We also identified candidate sets of BMP-activated TFs which, despite lacking known chondrogenic roles, have multiple lines of evidence supporting probable novel chondrogenic function. SMADs in this context also function almost exclusively in collaboration with SOX9, whereas SOX9 can operate independently from SMADs. We proposed a model where SOX9 is the primary regulatory signal in specifying chondrocyte cell identity during BMP-activated chondrogenesis, whereas BMP signaling cooperates with SOX9 to establish a permissive environment for chondrogenesis. Finally, we examined how the target genes of SMADs, SOX9, and the osteogenic TF RUNX2 changes during late endochondral ossification. We found that all three regulatory inputs cooperate to mediate a switch from a chondrogenic to osteogenic program, and that BMP signaling is now capable of establishing osteoblast identity independently from SOX9 and RUNX2. Many new target genes for all three TFs were identified, providing a novel resource to interrogate the changing GRN during endochondral ossification.Medicine, Faculty ofCellular and Physiological Sciences, Department ofGraduat