21793 research outputs found
Sort by
Phenomenology of Beyond Standard Model at Present and Future Colliders
No full textThis thesis investigates Beyond Standard Model (BSM) physics, focusing on three interconnected studies addressing fundamental puzzles in the Standard Model (SM) of particle physics. First, we examine the supersymmetric U(1)R × U(1)B−L model with both universal and non-universal boundary conditions at the GUT scale, studying its implications for dark matter, the muon g − 2 anomaly, and collider signals. Detailed reconstruction of final-state leptons establishes benchmarks with significance levels exceeding 5σ at the High-Luminosity LHC (HL-LHC). Second, we explore vectorlike leptons (VLLs) as extensions to the SM, demonstrating their potential to resolve both electron and muon anomalous magnetic moment discrepancies while consistent with the neutrino data. Six-lepton signatures can be clearly distinguished from SM
backgrounds at future hadron colliders operating at 100 TeV. Finally, we establish improved sensitivity to type-I seesaw superheavy Majorana neutrinos at future muon colliders in vector boson fusion (VBF). We show that μ+μ− colliders at 10 TeV and 10 ab^−1 integrated luminosity can exclude heavy Majorana neutrinos with mixing parameters |VμN| down to 10^−3 for masses up to 100 TeV. Altogether, these three studies target major sectors of BSM phenomenology through detailed present and future collider projections, providing compelling motivation for future experiments
Not All Green Is Equal: A Text-Based Analysis of Green Bond Yield Discount
No full textThis paper investigates the pricing dynamics of U.S. corporate green bonds issued between 2015 and 2024. While prior literature has mixed findings but often failed to identify a statistically significant greenium—the yield discount associated with green bonds—at issuance, my results reveal a weakly significant greenium in the primary market. More importantly, I document a consistently and highly significant greenium in the secondary market, suggesting that investors increasingly recognize and price in the environmental value of green bonds over time. The novel contribution of this study lies in the construction of a comprehensive green commitment score derived from the textual analysis of bond prospectuses. By categorizing and scoring green bonds based on their disclosed use of proceeds, project evaluation, external review, and reporting/transparency, where use of proceeds and project evaluation are the most significant, I show that bonds with stronger green commitments are associated with larger yield discounts in the secondary market. On the contrary, weak disclosure is associated with no green premium. These findings highlight the growing importance of transparency and credibility in green finance and underscore the role of disclosure quality in shaping bond pricing beyond the primary market
SWin: A Sliding Window Summarization Approach for Coherent LLM-driven Dialogue Systems
No full textLarge Language Models (LLMs) have revolutionized artificial intelligence (AI) driven dialogue systems, enabling various applications through advanced conversational capabilities. However, their stateless design impedes the development of sustainable assistants capable of evolving through extended interactions. Current memory mechanisms suffer from cascading error accumulation, architectural complexity, and computational inefficiency, which hinder scalability. To address this, we propose a novel sliding window-based memory module that dynamically updates the memory state using overlapping conversation windows, state-of-the-art prompt engineering techniques, and efficient summarization. The module balances contextual continuity with computational efficiency, reducing redundant token processing. This approach enhances the functionality of current mechanisms, and thorough tests on the Multi-Session Chat (MSC) dataset demonstrate that our approach is a reliable solution that exceeds automatic metrics produced by earlier approaches while optimizing token consumption. Our module's efficient design provides a workable solution for environments with limited resources, allowing dialogue systems to improve user interactions. Code, datasets, and evaluation scripts will be open-sourced to facilitate reproducibility and further research
Theoretical Investigation of Small Molecules Interactions with Metal-Organic Frameworks and Silicene
No full textUnderstanding and controlling surface–adsorbate interactions is crucial for designing advanced materials for applications such as toxic gas capture, sensing, and electronics. Although metal-organic frameworks (MOFs) are promising nanomaterial sorbents for toxic gases, the relationship between their structural and electronic properties and adsorption performance has not been established, limiting their rational design. Similarly, a detailed understanding of surface reactivity is lacking for other nanomaterials, such as silicene. Recent experiments show that silicene synthesized on highly oriented pyrolytic graphite (HOPG) exhibits stability against oxidation after weeks of air exposure, in contrast to previous results on other substrates and theoretical predictions. This discrepancy highlights the need for computational investigations to clarify surface reactivity and substrate effects. In this study, spin-polarized density functional theory (DFT) with Hubbard U correction is employed to investigate the adsorption mechanisms of toxic molecules on MOFs, focusing on M-MOF-74 (M = Mg, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn) and water-stable frameworks such as UiO-66, UiO-66-NH₂, UiO-67, MIL-53 and MFM-300. Toxic molecules adsorb on M-MOF-74 at unsaturated metals, whereas water-stable MOFs exhibit weak physisorption through hydrogen bonding. Including the Hubbard U correction is essential for accurately modeling the electronic structures and binding energies of MOFs containing transition metals. The M-MOF-74s display diverse magnetic behaviors. Interestingly, magnetic configurations do not significantly affect binding energies, suggesting that DFT calculations without considering magnetic states can reliably predict adsorption energies for these MOFs. However, cases such as V-MOF-74 show potential for magnetic sensing of NO₂. Furthermore, functionalization of MOFs, such as UiO-66-NH₂, enhances adsorption of NH₃, NO₂, and SO₂, highlighting the role of linker modifications in improving adsorption performance. As for silicene, DFT calculations using the generalized gradient approximation (GGA) indicate that both free-standing silicene and silicene on HOPG interact and react with O₂ via a barrierless process. However, DFT with hybrid functional reveals small energy barriers for oxidation, and Hartree–Fock methods predict significantly larger barriers, underscoring the sensitivity of predicted oxidation pathways to the choice of computational method. This study demonstrates the importance of advanced computational approaches and material design strategies in understanding the surface reactivity of materials
Black Masculinities in Canada: Reframing Black Men and Gender Advocacy
No full textThis dissertation examines how Black men in Canada experience and resist gendered racism, the intersection of racial and gender-based marginalization that shapes systemic and interpersonal barriers across the sexuality spectrum. Using a grounded theory design, this study draws on symbolic interactionism, Black feminism, and critical/decolonial masculinity studies to develop a humanizing account of Black male identity and community leadership.
Black masculinity is often represented from an American context and through a problem-centred framework within Western theory, particularly through uncritical applications of Hegemonic Masculinity Theory (HMT). These frameworks portray Black men as engaging in hypermasculine behaviours to compensate for societal constraints and oppression, reinforcing deficit-based and racialized narratives. In the Canadian context, where Black men face disproportionately high rates of incarceration, school dropout and streamlining, underemployment, and premature mortality, such framings obscure their lived realities and potential for social transformation.
Through semi-structured interviews with (n=21) Black male community organizers in Montreal, Ottawa, and Toronto, this research addressed three questions: (1) How do race and gender intersect to shape the lived experiences of Black men in Canada? (2) How do Black men construct and define their masculinities, and in what ways do their definitions challenge or reinforce dominant gender discourses? (3) How do Black male community organizers mobilize against gendered racism, and how do they contribute to community advancement in Canada?
Findings reveal that Black men encounter gendered racism through media stereotypes, educational marginalization, and workplace discrimination. However, they resist through two forms of intersectional gender advocacy: constructing counternarratives of masculinity and leading community-based efforts that address both gender and racial inequality. From these findings, the
ivstudy concludes by proposing Intimate Masculinities Theory (IMT), a grounded gender theory that centers emotions, vulnerability, and inner life as vital to Black male identity, expanding beyond the limits of power- and dominance-based models.
The study reframes Black men as a gender-equality deserving group in Canada, calling for greater recognition of their gendered experiences and advocacy work. It highlights their alignment with Black feminist approaches and underscores the need for gender theory to engage more deeply with participant narratives.
Advancing gender justice for Black men enhances well-being, supports cross-gender coalition-building, and offers a transformative framework for equity in education, employment, and media. Dissemination will include academic publications, a public event on Black men and gender justice, and a policy brief addressing gendered racism in Canadian schools, workplaces, and media, informed directly by participant recommendations
Journalism for a sustainable future - Defining the principles and exploring the role of human centric stories
No full textSustainability is a multifaceted concept that encompasses the interdependent dimensions of environmental stewardship, economic development, and social equity. Effectively communicating this complexity to the public requires nuanced, context-sensitive storytelling that accounts for historical actions, current consequences, and future mitigation strategies. However, sustainability often does not align easily with conventional journalistic practices. This research-creation thesis developed and tested a pragmatic guide aimed at integrating sustainability principles into journalistic work. The guide is informed by participatory action research and a qualitative survey conducted with members of Concordia University’s sustainability community. These insights helped identify key elements necessary for meaningful communication about sustainability in journalism. The guide was then applied in the field through the creation of audio and visual stories featuring two Concordia initiatives: Cultivation and Hive Free Meals.
While Bonfadelli (2010) observes that sustainable journalism remains marginal due to the often invisible nature of sustainability issues, this thesis argues that many tangible, solution-oriented initiatives exist and deserve greater visibility. Journalists have a responsibility to bring these efforts to the forefront, enabling public engagement and collective action. Our future depends on present decisions, and through intentional storytelling, journalism can play a transformative role in shaping a more sustainable society
Identity negotiation in digital environments: A case study of L2 influencers
No full textIdentity negotiation has been studied as an important aspect of second language (L2) learning especially within Darvin and Norton’s (2016) expanded model of investment which help describe sites of conflict amongst identity, ideology and capital of both the learner and the communities they encounter which include digital spaces. While studies about L2 online identity negotiation have used this model to study how L2 speakers negotiate identity as learners, there is a need to study L2 speakers who present themselves in the digital space outside of the label L2 learner. In this study I interview two social media influencers who described their journey through identity negotiation on social media platforms. Both participants described various points of conflict that were consistent with how identity negotiation functions in Darvin and Norton’s expanded model of identity in order to position themselves as legitimate speakers in their context as social media influencers. The results from this study affirm the ways social media allows a place for L2 speakers to use their language outside of the L2 or L2 learner identit
Social Media and L2 Pragmatic Awareness: Insights from Instagram
No full textOne of the challenging aspects of teaching in English as a Foreign Language (EFL) context has always been second language (L2) pragmatic instruction since learners have fewer opportunities to use English in their interactions outside the classroom and textbooks do not fully address L2 pragmatics. This study focused on participants’ perception of task-based weekly lessons using Instagram aimed to raise their pragmatic awareness. Activities were designed according to Taguchi’s (2017) criteria for task design to help participants analyze pragmatic features in Instagram videos on different topics: book reviews, discussions and professional conversations. Over the course of 1 month, 4 participants with an EFL background, two of whom now reside in an English-speaking country, took these weekly lessons and wrote in reflection journals after completing the tasks. Participants noticed and analyzed a range of pragmatic features even though learning about pragmatics was new to them. They also shared which strategies helped them make these activities easier. Participants reported a more positive perspective on using Instagram for L2 pragmatics in addition to noting a few of strategies used during the study. This study highlights the value of Instagram as a learning tool, role of teachers in teaching pragmatics, and the need to reflect on strategies when using Instagram to enhance pragmatic awareness
A Semi-Automated, High Throughput Method for Genetically Engineering and Phenotyping Yeast Extracellular Vesicles
No full textExtracellular vesicles (EVs) represent a promising new modality for drug delivery. We use Saccharomyces cerevisiae (baker’s yeast) – an organism used for drug biomanufacturing – as a platform to design, build, and test engineered EVs for therapeutic applications. This involves modifying their contents and surfaces by adding human or yeast proteins with diverse functionalities requiring testing of thousands of modifications (individually or in combination) to optimize EV cargo loading, cell targeting, and content delivery tailoring to specific outcomes in patients. To support our studies, I sought to establish high throughput cloning and phenotyping protocols to generate libraries of genetically modified S. cerevisiae strains. I employed Golden Gate and Gateway cloning strategies based on the modular Yeast Toolkit, enabling use of new constructs by the synthetic biology community. Candidate genes were introduced into donor plasmids and then integrated into expression vectors containing a strong promotor (TDH3) and the Nanoluciferase (NLuc) gene. PCR, genetic assemblies, bacterial transformations and colony selection were conducted in 96-well plate format by robotic equipment housed in Concordia University’s Genome Foundry. After assembled vectors were validated by pooled nanopore sequencing, robots were used to transform them into S. cerevisiae (e.g. wild type BY4741), to select clonal transformants, and to prepare furizamine-based assays for detection of candidate proteins tagged with the luminescent biomarker nanoluciferase (NLuc) within whole cell lysates or EV-containing samples, measured using a plate-reading luminometer. As proof-of-concept, I implemented an automated procedure to generate an initial set of 96 yeast strains each expressing a candidate protein fused to (NLuc). Despite using a single promotor, I observed variable expression levels of human and yeast candidate proteins within S. cerevisiae cells. Initial phenotyping of extracellular media containing EVs revealed the presence of some protein candidates, later confirmed by assessing EVs purified by ultrafiltration and size exclusion chromatography. These included human proteins (e.g. CD81) suggesting that the mechanism(s) underlying EV protein loading are conserved. In all, I developed an automated yeast genetic engineering and phenotyping research pipeline for high throughput screening of strategies to improve EV functionalities required for use as next-gen drug delivery vehicles
Auxiliary llearning for patch and WSI pathology image classification
No full textEarly and accurate cancer detection through pathology imaging plays a critical role in improving patient outcomes. Despite promising advances from deep learning (DL) models, their real-world deployment is hindered by significant challenges.
To address the performance degradation caused by domain shifts—variations from different imaging devices, staining protocols, and patient demographics—we introduce PathTTT. This novel framework enhances model robustness at the patch level by combining Test-Time Training (TTT) with Model-Agnostic Meta-Learning (MAML). This bi-level optimization strategy leverages MAML to train a model for rapid adaptation, while TTT dynamically fine-tunes its parameters during inference, allowing for effective generalization to unseen distributions.
A separate, key challenge in computational pathology is the effective analysis of whole slide images (WSIs). While multi-instance learning (MIL) has become a widely adopted paradigm for WSI classification, it often suffers from overfitting due to the extremely high dimensionality and heterogeneity of WSIs, combined with the limited availability of annotated data. To address these challenges, we propose Masked Feature Embedding for Multi-Instance Learning (MFE-MIL), an innovative method designed to enhance existing end-to-end MIL pipelines. By introducing a masked feature embedding prediction task, our method provides a strong self-supervised signal that forces the model to learn highly discriminative and context-aware representations from instance features.
Extensive experiments on multiple benchmark pathology imaging datasets demonstrate that both PathTTT and MFE-MIL consistently outperform state-of-the-art methods in their respective domains. These results collectively underscore the potential of this work to facilitate more reliable and generalizable cancer detection systems in real-world clinical applications