21793 research outputs found
Sort by
Advanced Blind Source Separation Methods for Multivariate Data Modeling and Clustering
Get PDFAs the amount of data being generated keeps growing, there is an increasing demand for adaptable approaches that can effectively extract the overall trends while also maintaining subject-specific information from large-scale datasets. Modeling complex and high-dimensional data presents significant challenges across healthcare, human action recognition, and speech recognition.
In this dissertation, we develop a bounded multivariate generalized Gaussian mixture model (BMGGMM) integrated with independent component analysis (ICA) to effectively capture the correlated features in multivariate data. While independent vector analysis (IVA) extends ICA to handle multiple datasets by leveraging inter-dataset dependencies and preserving their correlation structures, it suffers from limitations when dealing with complex datasets. To overcome this, we propose a novel blind source separation (BSS) method that combines IVA with the BMGGMM framework, enabling robust modeling of complex data distributions with varying shapes and dimensions. Second, we introduced the integration of the ICA-BMGGMM and IVA-BMGGMM to the hidden Markov model (HMM) to boost their performance in terms of source separation. The performance of IVA deteriorates as the number of datasets and sources increases. To address this limitation, we propose the adaptive constrained IVA (acIVAMGGMM) and bounded acIVAMGGMM techniques. These methods integrate multiple reference signals into the IVA function and adaptively control the reference-estimated source relations. Finally, we introduce a new approach, ICA and IVA for common subspace analysis (ICABMGGMM-CS) and IVABMGGMM-CS, designed for the subspace analysis of multi-subject fMRI data. These methods leverage the strengths of both ICA and IVA while effectively addressing the challenges posed by high dimensions
Evaluating Website Data Leaks through Spam Collection on Honeypots
Get PDFPeople increasingly rely on online services for communication, education, shopping, entertainment
etc., but this convenience comes with escalating spam volumes. Prior studies have linked spam primarily
to user behavior or data leaks but lacked effective methods to distinguish these causes. Our research
proposes using spam as a forensic indicator of data leaks and introduces a low-complexity fingerprinting
technique to trace leak sources, while evaluating consent and subscription practices of websites and
telecom.
We deployed 148 honeypots with 740 accounts across 370 websites in 12 communities over 12
months, analyzing 12,490 spam emails to assess forensic indicators and evaluate exposure, privacy policy
exploitation seen with legitimate websites under the Canadian Anti-Spam Law (CASL). Our method
was bench-marked against traditional leak detection models and fingerprinting techniques.
Our findings reveal that many legitimate websites exploit CASL consent practices by automatically
enrolling users in mailing lists via implicit consent, while sites requiring explicit consent often violate
their own policies, highlighting enforcement gaps. We recommend that regulators mandate a clear separation
between subscription agreements and privacy policies and require explicit third-party consent at
sign-up.
Additionally, our analysis shows that men aged 48–57 receive the highest spam volumes, with peak
activity between 00-04 minute of each hour, with peaks on Thursdays. These insights offer valuable
guidance for enhancing spam filtering models.
Using our analysis engine, we achieved 100% leak detection and 99.29% source attribution accuracy.
Compared to network intrusion detection, log analysis, machine learning, and traditional fingerprinting,
our method more effectively identifies compliance violations, traces leaks to their source, and estimates
exposure impact. Furthermore, to address telecom issues from phone number reassignments, we proposed
a nonce-based de-association method that promises significant spam reduction
My Mother Always Said, “Don’t Be Lazy With Your Inuktitut: Dismantling Linguistic and Cultural Imperialism in Nunavik.
No full textIn a globalized world where English language and Western culture dominate, Nunavik faces a growing threat of language and cultural loss. This dissertation responds to the urgent need for Inuit-led resistance against colonial pressures by fostering self-determination and cultural resilience in education. Through three interconnected manuscripts, it addresses this need by laying a foundation for decolonizing English second-language (ESL) education in Nunavik. Manuscript A critically examines Nunavik’s colonial history in language education and outlines the importance of Inuit Qaujimajatuqangit as a framework for decolonizing ESL. Manuscript B uncovers gaps in intercultural communication and social media use for protecting Indigenous language and culture, underscoring the lack of Inuit-specific research. These findings inform the empirical study in Manuscript C, which amplifies Inuit voices on how to protect Inuktitut and culture. This dissertation culminates in the development of Silattusarniq, an Inuit-specific methodological framework for research that is grounded in Inuit values and perspectives
Of Tables, Tongues, and Pineapples: Visualising the Sense of Taste in Early Modern Dutch Still Life and Natural Philosophy
No full textIn The Art of Painting (Amsterdam 1707), the Dutch artist and writer Gérard de Lairesse (1641-1711) stated that the sense of taste “cannot be painted ; but may be in some measure expressed by occult significations.” During the early modern period, European subjectivities came in contact with “new” foods and flavours that they had not experienced before. In order to make these experiences intelligible, they embarked on a journey for visualising the invisible. This dissertation explores the multiple strategies adopted by Dutch artists and natural philosophers to visualise the sense of taste.
Drawing from the fields of sensory studies, and the interconnected histories of art and knowledge, this project aims at investigating the role of the sense of taste in late seventeenth-century and early eighteenth-century Dutch art from a global perspective. It explores early modern ideas about taste and sense perception, and their role in art theory and scientific discourses on nature and the body. Modelling subjectivity and shaping society, the sense of taste was conceived within a larger project of nation-building and world-making.
Looking across a variety of visual genres and media, this dissertation proposes an iconography of the sense of taste that encompasses allegories of the five senses, still life, and scientific illustration. It looks at these images in light of an array of textual sources, including art theory, travel literature, natural philosophy, and culinary recipes. In order to do so, it considers three different case studies: Gérard de Lairesse’s views on still life and the five senses; the tongue in anatomical illustration; and, finally, the global history of the pineapple as the epitome of taste and the ineffable
Pretty in Plastic: Drawing Sustenance from the Capitalist Wasteland in Portia Munson’s Pink Projects
Get PDFThis thesis examines the thirty-year installation and sculpture practice of New York-based artist Portia Munson (b. 1961, Beverly, MA) involving the collection and assemblage of discarded plastic objects of the same colour, specifically the Pink Project series. While her assemblage works of blue and green plastic trash present more obvious ties to plastic consumption and its detrimental impact on the environment, the Pink Projects series reveals the scale of the pink palette of products that become consumer waste, and enables an understanding of both the work’s subject and materiality as the abject product of gendered capitalist ideology. I consider four of Munson’s works that assemble pink plastic waste. This thesis argues that Munson’s work critically visualizes the paralyzing, overwhelming, and terrifying nature of our planetary predicament, while also enacting a feminist reparative ritual — drawing from the history of women’s altar-making, a feminist history of capitalism, eco-criticism, anti-capitalist philosophies, and depression studies. The transformative quality of Munson’s works hinge on their expression of the degraded and abject status–produced in the transition to and maintenance of capitalism–of waste and female body as productive spaces through which the unknown may be encountered and therefore from which the critical vernacular to address planetary chaos is produced
Digital Deceptions: Unveiling the Impact of Pseudo-Local News on Democracy and Crafting Countermeasures (Metric Media Case Study)
Get PDFMetric Media is known for its vast network of US-based pseudo-local news sites propagating specific ideologies, operating beyond traditional mass media manipulation techniques. By micro-targeting societal segments on divisive issues, Metric Media presents a nuanced form of content dissemination, raising questions about its impact on public opinion and welfare. The case study employed sentiment analysis, topic modelling, and social media engagement metrics to analyze content from July 2019 to January 2024. It offered a comprehensive view of Metric Media's thematic focus and the resonance of its narratives among communities. Additionally, it explored the role of data scraping and visualization in highlighting the organization's strategic approaches to news dissemination. This investigation not only elucidated Metric Media's use as a political tool across ideologies but also underscored the importance of data-driven methodologies in enhancing public awareness of media influence. The findings contribute to the discourse on digital media manipulation, emphasizing the need for critical engagement with news sources.
Keywords: local news, open source intelligence (OSINT), data journalism, propagand
Statistical Analysis of Energy Measures as Biomarkers of SARS-Covid-2 Variants and Receptors
Get PDFStatistical Analysis of Energy Measures as Biomarkers of SARS-Covid-2 Variants and Receptors
Khawla Ghannoum Al Chawaf
The COVID-19 outbreak has made it evident that the nature and behavior of SARS-CoV-2 require constant research and surveillance, owing to the high mutation rates that lead to variants. This work focuses on the statistical analysis of energy measures as biomarkers of SARS-CoV-2. Thus, three statistical tests are applied to the data: the multiple ANOVA test for equality of means, Bartlett’s test for equality of variances, and Levene’s test for assessing the homogeneity of variances. These tests aim to determine which energy measure can differentiate between SARS-CoV-2 variants, human cell receptors (GRP78 and ACE2), and their combinations.
To further investigate the specific pairwise differences between groups, Tukey’s HSD test was performed after the ANOVA. The Tukey test provided adjusted p-values (p-adj) for each pairwise comparison, allowing for a more detailed understanding of significant differences in energy measures across variants, receptors, and their combinations.
The proposed approach combines energy measures and sequence data to develop classification systems and brings out the variety of the virus’ genetics and interaction mechanisms. This work aims to improve the accuracy of variant identification and contribute to creating tailored interventions, which would help address the COVID-19 issue and contribute considerably to the global fight against the pandemic.
Keywords: SARS-CoV-2, COVID-19, statistical analysis, variant identification, human receptors, genetic sequences, ANOVA test, Bartlett’s test, Levene’s Test, Tukey’s HSD test
Do Degrees Still Pay Off? Changing Returns to Education in Canada from 2005 to 2020.
Get PDFThis paper studies how returns to educational degrees has evolved in Canada from 2005 to 2020 by combining data from the 2006 and 2021 Canadian Census. The study uses a two-class finite mixture model to capture unobserved heterogeneity that may be correlated with individual characteristics, in a way that resembles the use of fixed effects in a panel data setting. The mixture model reveals two distinct groups and highlights unobserved differences in earnings
that standard models could miss. The findings show that while higher education still leads to higher earnings, the return to education has empirically declined over time for individuals
with graduate degrees. In contrast, returns to college and trade-level education have increased, suggesting that practical skills has become more valuable in the Canadian job market. These findings reveal the changing nature of the Canadian economy and help policymakers and educators improve education and training programs to better meet the needs of today’s job market
Shaped by Society: Uncovering the Sociological Factors Behind the Development of Serial Killers
Get PDFSerial killing is one of the most extreme and rare forms of deviant behaviour, but much of the existing research remains focused on psychological and criminological explanations. This critical literature review examines serial murder through a sociological lens, identifying key risk factors that contribute to the development and actions of serial killers. By emphasizing broader social, environmental, and structural influences, this essay challenges the notion that serial murder is solely an individual pathology. It explores how factors such as childhood abuse, dysfunctional family dynamics, socio-economic marginalization, media influence, and violent subcultures interact to shape serial killers’ trajectories. The analysis draws on both positivist and social constructionist perspectives. Findings indicate that childhood trauma, particularly abuse, is a recurring risk factor among serial killers, though not a universal one. Additionally, media portrayals can contribute to the construction of serial killers as cultural figures, influencing both public perception and offenders’ self-identities. Contrary to popular belief, socio-economic status alone does not appear to be a primary determinant of serial murder. By synthesizing various sociological perspectives, this review highlights the intersection of social forces that contribute to serial killing, positioning it as a product of broader structural and cultural dynamics within society rather than an isolated phenomenon
Critical Analysis of the Quebec Policy on Education
Get PDFThis critical policy analysis seeks to identify the underlying ideologies, as well as the government’s motivations and intentions, by examining the Policy on Educational Success (PES) (Gouvernement du Québec, 2017). The PES was developed to provide a framework for educational reforms from 2017 to 2030 in Quebec. This analysis considers the influence of globalization and the unique socio-political context of Quebec while utilizing Fairclough’s (2007) and Bacchi’s (2016) critical discourse analysis. The results indicate the PES mirrors many structural reform recommendations found in the Organization for Economic Co-operation and Development’s (OECD) education policy. This analysis finds that these reforms are integrated within the Quebec education system with two intentions. The first is to cultivate human capital to support and sustain economic development. The second and most prominent intention of the PES is to inculcate the population with a national identity, dominated by Quebec values of secularism and the French language. The conclusion of this thesis is a warning that the reform recommendations of the OECD, in the movement toward globalization, are creating systems of centralized authority, which are vulnerable to being co-opted by government powers and used for the indoctrination of the population. This analysis contributes to the body of research on education policy, specifically, on Quebec education policy, providing a much-needed critical perspective