21090 research outputs found
Sort by
Integrating Cognitive Work Analysis into an ACT-R Model for Cybersecurity Applications
No full textCybersecurity is a trending concern with the rapid development of many systems. While humans are often considered vulnerable targets, research on human factors remains limited compared to the extensive technical focus on defense and mitigation strategies. Human-focused cognitive research in this domain faces two primary challenges: the evolving and complex nature of the cybersecurity landscape, and the domain-specific characteristics of the systems under attack. These challenges point to the need for modeling human performance in identifying vulnerabilities, with both precise dynamic measurement and domain-specific fidelity.
Accordingly, we proposed a solution by integrating CWA into ACT-R models. A detailed elaboration on the CWA and ACT-R's structural compatibility across dimensions, their fundamental strengths as complements, and the functional competencies with integration was presented. This conceptual exploration demonstrated the feasibility of integrating the CWA and ACT-R, leading to improvements in model construction efficiency and domain-specific validity.
We explored CWA and ACT-R for modeling humans in vehicle cybersecurity. While we were able to demonstrate a model, a follow-up study with human participants showed that drivers may not actively identify vulnerabilities and mitigate cyber threats. We then practically implemented and applied the integrated model, from model construction preparation to detailed rule development, guided by CWA’s Work Domain Analysis, Control Task Analysis, and Strategies Analysis, to simulate the SOC analysts' cybersecurity alert triage performance. The model construction process demonstrated better efficiency with a systematic approach, and the resulting model showed improvement trend in quantitative accuracy, domain-specific validity, and the interpretability of human adaptability and flexibility. However, the model is limited in capturing human exploratory behavior, prompting a brief test of using Generative AI (GAI) models to address this gap.
This thesis is the first exploration and implementation of integrating CWA-guided domain-specific analysis with ACT-R’s computational capabilities to develop an integrated cognitive model for humans in complex work domains. The effort advances the development of cognitive modeling by providing theoretical grounding and practical insights for applying and extending cognitive models. Finally, we discuss whether GAI models might enhance cognitive modeling, as GAI capabilities become more available
Woodland ecosystem services of the past and present in Herstmonceux and South England
No full textThis thesis quantifies ecosystem services from broad- and mixed-leaved woodlands in the southern United Kingdom. Synthesizing concepts and approaches from historical ecology, environmental history, community ecology, dendrological allometry, and punctuated equilibrium theory, this thesis illuminates the multifaceted ways in which woodlands provide ecosystem services that are valued by people in the UK, both socially and ecologically. A complex relationship between people and woodlands emerges through time, which can be understood directly via management decisions, but also more abstractly through the sentiments that people attach to trees. Both of these approaches carry normative implications about the value of particular woodland ecosystem services. While the values that guided past decisions in woodland management are not always explicit, archival maps and remotely sensed data can reveal the nature of land use changes that manifest over long periods of time, i.e., 150 years. Within a case study context in Herstmonceux, East Sussex, archival data demonstrated a progression towards a modern day multifunctional wooded landscape.
Within this modern context, historical woodland management regimes like coppicing drive specific ecosystem services like biodiversity and carbon storage to change measurably on a near-annual basis. This indicates that historic management regimes have important implications for ecosystem service provision not just over the course of generations, but also on fairly short time frames, e.g., 15 years. Land managers of coppice woodlands must therefore be cognizant of how everyday land management decisions can impact the ecosystem services, and therefore values (both intrinsic and instrumental) derived from them. Importantly, land management decisions and regimes also change abruptly in response to exogenous factors. When extreme damage was caused to trees and woodlands as a result of the October 1987 Great Storm, there occurred a traceable punctuation reflected in both the public sentiment and the priorities of woodland managers regarding trees. Changes in woodland ecosystem services can thus be slow-moving or sudden. Ultimately, it is this complex, always-changing relationship between humans and the environment that shape not only the actual provision of ecosystem services, but also perceptions of that provision, and, furthermore, how ecosystem services themselves are valued.
The ecosystem services perspective, therefore, may be applied to and represent both intrinsic and instrumental values, rather than solely instrumental values, which has been a longstanding critique of the framework. However, researchers aiming to employ the ecosystem services framework in this manner must be intentional and explicit in their doing so, in order to shift the guiding paradigms in conservation away from “nature for people,” and towards a “people are nature” perspective
Demystifying Foreground-Background Memorization in Diffusion Models
No full textDiffusion models (DMs) memorize training images and can reproduce near-duplicates during generation. Current detection methods identify verbatim memorization but fail to
capture two critical aspects: quantifying partial memorization occurring in small image regions, and memorization patterns beyond specific prompt-image pairs. To address
these limitations, we propose Foreground Background Memorization (FB-Mem), a novel
segmentation-based metric that classifies and quantifies memorized regions within generated images. Our method reveals that memorization is more pervasive than previously
understood: (1) individual generations from single prompts may be linked to clusters of
similar training images, revealing complex memorization patterns that extend beyond one-to-one correspondences; and (2) existing model-level mitigation methods, such as neuron
deactivation and pruning, fail to eliminate local memorization, which persists particularly
in foreground regions. Our work establishes an effective framework for measuring memorization in diffusion models, demonstrates the inadequacy of current mitigation approaches,
and proposes a stronger mitigation method using a clustering approach
Green Space Equity & Environmental Justice: A Comparative Study between North St. James Town & High Park-Swansea Communities in Toronto
No full textIn cities shaped by unequal development and growing environmental pressures, urban green spaces are increasingly recognized not just as aesthetic luxuries but as fundamental components of livable, healthy, and equitable communities. While public parks and naturalized areas linked to a wide range of social, mental, and ecological benefits, the accessibility and distribution of these spaces are often influenced by systemic inequities embedded in urban planning, land use policy, and neighborhood development trajectories. This thesis investigates these disparities by comparing two socioeconomically and spatially distinct communities, High Park-Swansea (HPS) and North St. James Town (NSJ). Guided by social‑ecological systems (SES), political ecology, and community‑based participatory research (CBPR), the study asks how social fabric and planning histories shape the equitable distribution and lived experience of public green space.
The research investigates green space quality, accessibility, and use in each neighborhood, focusing on how socio‑economic status, density, infrastructure, and community engagement intersect to produce divergent relationships with urban nature. Drawing on 63 surveys and 24 in‑depth interviews, it employs inductive coding and narrative analysis to identify themes related to accessibility, inclusiveness, safety, ecological quality, and psychological well‑being. Findings show that access to green space is not defined by proximity or quantity alone but is closely tied to perceptions of safety, historical marginalization, and belonging.
HPS emerges as a neighborhood with relatively high green space coverage, affluent demographics, and strong stewardship, while NSJ is characterized by dense high‑rise housing, constrained green infrastructure, and heightened social vulnerability. An analysis of Toronto’s green space policies indicates that comprehensive goals are often undermined by weak enforcement, a lack of spatially disaggregated values, and limited community‑oriented design standards, contributing to a spatial politics of exclusion. Moreover, it explores how Toronto’s green space policies, while comprehensive on paper, often lack enforcement mechanisms, spatially disaggregated benchmarks, and community-oriented design standards.
By mapping community narratives to broader structural trends, the study reveals how planning practices, past and present, contribute to a spatial politics of exclusion where certain communities are underserved by design. In doing so, it informs concrete recommendations for municipal planners, including the development of neighborhood-level green space equity indicators, integration of community-informed design criteria in development approvals, and policy tools that ensure green infrastructure investment is responsive to local needs. These insights hold relevance beyond the context of Toronto, contributing to global conversations on urban sustainability, environmental justice, and inclusive planning decisions
Classification Results for Intersective Polynomials With No Integral Roots
No full textIn this thesis, we algebraically classify strongly intersective polynomials - polynomials with no integer roots but with a root modulo every positive integer - of degree 5--10. In particular, we compute a list of possible Galois groups of such polynomials. We also prove constraints on the splitting behaviour of ramified primes (i.e. primes that ramify in a splitting field of the polynomial). In the process, we show that intersectivity can be thought of as a property of a Galois number field, together with its set of subfields of specified degrees. This was achieved with characterisations of Berend-Bilu and Sonn, the latter of which we also generalise. Implementations in SageMath and GAP are provided. We also utilise Hensel's Lemma and other standard results on the local behaviour of simple field extensions
The necessity of motoric engagement in enhancingroute memory
No full textThe relative contribution of decision-making and motor engagement at encoding, on route memory, was examined using virtual reality (VR). During encoding, participants explored 12 virtual environments for 40 s each. Navigation strategy during encoding was manipulated within-subjects. On Active trials, participants made decisions about their route of travel. On Guided trials, they followed a pre-determined path overlaid on the road, removing the need for decision-making. On Passive trials, participants sim-ply viewed a set route, without initiating decision-making nor engaging movement during encoding. Following exploration of each environment, participants were asked to ‘re-trace their steps’ using the exact route they had just travelled. We also manipulated type of VR implementation(Desktop VR, Headset VR) between subjects. Movement in a Desktop-VR group was controlled via keyboard input, limiting motoric engagement. Movement in a Headset-VR group occurred using a VR-compatible steering wheel, re-quiring relatively greater motoric engagement. We found an effect of navigation strategy only in the Headset-VR group:route memory was significantly better following Active and Guided relative to Passive trials. Memory did not differ following Active relative to Guided trial types, suggesting that decision-making does not underlie the memory benefit. We suggest route memory is enhanced when initiating physical movement during encoding.NSER
Revisiting the ‘Lensing is Low’ Problem With UNIONS
No full textIn this thesis, we present new measurements of the galaxy–galaxy lensing (GGL) signal around Baryon Oscillation Spectroscopic Survey (BOSS) CMASS galaxies using background sources from the Ultraviolet Near-Infrared Optical Northern Survey (UNIONS). With an overlap of approximately 2650 square degrees between CMASS lenses and background source galaxies—the largest to date—we obtain precise large-scale GGL measurements. With these new measurements, we revisit the so-called ‘lensing is low’ problem, wherein galaxy–halo connection models calibrated on clustering data over-predict the GGL signal by 20–40% under cosmic microwave background (CMB)-based cosmologies. We model the galaxy–halo connection using a halo occupation distribution (HOD), and perform joint fits to both GGL and clustering signals across a wide range of scales, as well as a clustering-only fit. Similar to previous work, we find a lensing–is–low effect in the CMASS sample, although our GGL and clustering predictions are less inconsistent with each other. The best joint fits are achieved by lowering the amplitude of the matter power spectrum relative to Planck 2018, driven by the precision of our large-scale GGL measurements. Once a lower matter power spectrum amplitude is adopted, feedback is the only HOD extension that further improves the joint fit. Our feedback model redistributes matter within a halo, modifying the halo–matter cross–power spectrum. Overall, we find that two models describe our observables equally well: one where HOD and cosmological parameters are free, and one where HOD, cosmological, and feedback parameters are free. Importantly, we emphasize the role of large scales in driving the lensing–is–low effect, shifting the narrative away from a purely small-scale issue
Investigating the Performance of Straight and Bent GFRP Bars as Flexural Reinforcement for Glulam Beams
No full textThe heightened interest in using wood as a sustainable building material contributed to an increased demand for glued-laminated timber (glulam). Despite this, fundamental research is required on how to rehabilitate and retrofit deficient structural wooden members to extend the service life of the structure. The research focuses on the effects of reinforcement configurations consisting of glass-fibre reinforced polymer (GFRP) bars on the flexural behaviour of glulam beams. Of particular interest are the effects of reinforcement length, adhesive type, and knurling on the failure modes of the reinforced members when compared to unreinforced glulam. A total of eighteen pullout specimens were tested to investigate the effects of adhesive and knurling patterns on bond strength, and fourteen full-scale glulam beams were tested to failure under four-point static bending, including four unreinforced and ten GFRP-reinforced. The pullout test results showed that the texture and density of an adhesive had a critical role on the overall behaviour with improved behaviour in specimens using a fluid density in comparison to those with dense. The addition of GFRP reinforcement to the glulam beams contributed to an increase in strength, failure displacement, and stiffness by factors ranging between 1.16 – 1.30, 1.04 – 1.24, and 1.13 – 1.18, respectively, in comparison to unreinforced glulam irrespective of the failure mode obtained. The effects of reinforcement length and termination point showed that the change from short to long bars resulted in improvements in maximum resistance and stiffness by factors of 1.10 and 1.19, respectively, for the bent bar reinforced specimens, and insignificant improvements for the straight bar reinforced specimens. Additionally, the change from straight to bent bars resulted in improvements in maximum resistance and stiffness by factors of 1.06 and 1.03, respectively, for the specimens with longer lengths of bars, and insignificant improvements for the specimens with short lengths of bars. The addition of knurling in the full-scale GFRP-reinforced beams resulted in increases of 1.07 and 1.03 for the maximum resistance and stiffness, in comparison to beams without knurling. Additionally, a change in failure mode from shear to flexure was observed with the addition of knurling. A material model was developed to predict the flexural behaviour of unreinforced and GFRP-reinforced glulam beams, and the two proposed approaches were shown to generally captured the overall behaviour with a tendency to overpredict displacements at initial failure. Finally, the improvement in tensile failure strains in flexure due to the reinforcement was not observed to be present due to the mixed failure modes of shear and flexure. Strains from the digital image correlation system were observed to be lower than those measured by localized strain gauges, suggesting that measuring strains over a large area is critical
Controlled Degradation of Biodegradable Polymers for Use in Melt-Blown Nonwovens
No full textNonwovens have many applications; however, they are typically made from petroleum-based polymers that are unsustainable and release microplastics upon breakdown that pose a risk to the health of both us and the environment. As such, it is necessary to shift to more environmentally benign materials such as biodegradable polymers. Unfortunately, readily available biodegradable polymers do not have the properties required, namely high melt flow index (MFI), for use in melt blowing, a common method for producing nonwovens. Therefore, this thesis aims to modify biodegradable polymers through aqueous hydrogen peroxide-induced controlled degradation to make them compatible with the melt blowing process.
Two different biodegradable polymers are subjected to this treatment: poly(lactic acid) (PLA) and poly(butylene adipate-co-terephthalate) (PBAT). Both systems demonstrated clear evidence of molecular weight reduction due to random chain scission induced by the peroxide radicals with additional contributions from thermal degradation and hydrolysis. PLA was also shown to begin crosslinking once a critical processing time was reached, while processing time appeared to have little effect on PBAT. The degraded products were then melt-blown to produce nonwoven mats with much finer and more uniform fibers compared to their untreated counterparts. The degraded PBAT from the second study was then blended with untreated PLA to develop a final melt-blown nonwoven with more balanced tensile properties than either material alone. Finally, since PLA and PBAT are immiscible when both are present in large quantities, maleation was used as a compatibilization technique to successfully enhance the quality of the melt-blown blends. Overall, reactive batch mixing with aqueous hydrogen peroxide is demonstrated to be a sustainable method for the molecular weight reduction of low MFI biodegradable polymers, allowing them to be well suited for use in melt-blown nonwovens
Path integral and qubit encoding techniques for quantum simulations of discrete planar rotor lattices
No full textTypical path integral Monte Carlo approaches use the primitive approximation to compute the probability density for a given path. In this thesis, we investigate the utility of pair approximating the action in path integral ground state simulations targeting planar rotations. The pair propagator, which was initially introduced to study superfluidity in condensed Helium, is naturally well-suited for systems interacting with a pair-wise potential. Consequently, paths sampled using the pair action tend to be closer to the exact paths (compared to primitive Trotter paths) for such systems leading to convergence with less imaginary time steps. Our approach relies on using the pair factorization in conjunction with a rejection-free path integral ground state paradigm to study a chain of planar rotors interacting with a pair-wise dipole-dipole interaction. We first use a heat kernel expansion to analyze the asymptotics of the pair propagator in imaginary time. Then, we exhibit the utility of the pair factorization scheme via convergence studies comparing the pair and primitive propagators. Finally, we compute energetic and structural properties of this system including the orientational correlation and Binder ratio as functions of the coupling strength to examine the behavior of the pair-DVR method near criticality. Density matrix renormalization group calculations are used for benchmarking throughout.
Near term quantum devices have recently garnered significant interest as promising candidates for investigating difficult-to-probe regimes in many-body physics. To this end, various qubit encoding schemes targeting second quantized Hamiltonians have been proposed and optimized. In this thesis, we also investigate two qubit representations of the planar rotor lattice Hamiltonian. The first representation is realized by decomposing the rotor Hamiltonian projectors in binary and mapping them to spin-1/2 projectors. The second approach relies on embedding the planar rotor lattice Hilbert space in a larger space and recovering the relevant qubit encoded system as a quotient space projecting down to the physical degrees of freedom. This is typically called the unary mapping and is used for bosonic systems. We establish the veracity of the two encoding approaches using sparse diagonalization on small chains and discuss quantum phase estimation resource requirements to simulate small planar rotor lattices on near-term quantum devices