21793 research outputs found
Sort by
One Species amongst Many: Creatively Thinking through Anthropocentrism
Full text linkBy analyzing the work of three contemporary artists, this thesis intellectually grapples with how different mechanisms of interspecies attunement can disentangle, challenge, and in some instances reiterate, modalities of anthropocentric thought and action. The first section examines Špela Petrič’s performance piece Skotopoiesis (2015), which seeks to reconfigure human-plant relationalities and engender expanded anthropic engagement with the vegetative world. Petrič’s work provides a generative ground for contending with modes of attunement such as lengthened temporality, cross-species communication, and the formation of meaning. The second part discusses Tomás Saraceno’s interactive installation Play-Ground (2024), which aspires to bridge the sensorial worlds of humans and arachnids. Play-Ground generates a space to wrestle with notions of imaginative and embodied perception, while exposing the difficulty of thinking beyond anthropomorphism. The third section teases apart Alexandra Daisy Ginsberg’s project that prioritizes the specific niches and needs of pollinators. Entitled Pollinator Pathmaker (2021-ongoing), Ginsberg’s endeavour demonstrates how efforts of creative empathy and multispecies cooperation can instantiate ecologies of reciprocity, ethical frameworks of care, and multi-species coexistence. This thesis not only investigates how artistic praxis engages with paradigms of anthropocentrism, but actively confronts how art can employ creative methods to (re)construct our relationship with non-human entities. With the continual intensification of ongoing ecological imperatives, such work is integral as it highlights the challenges, as well as the value, of employing artistic endeavours to support interspecies flourishing
Weighted Federated Averaging in Verifying Sensor Reading in Smart Homes to Mitigate Malicious Attacks
No full textFederated learning (FL) in event verification of smart homes increases the accuracy of any events, e.g., door opening. If the global model in federated learning gets poisoned, the verification of the event for all the participants will potentially decrease. Existing solutions a. Not considering federated learning in event verification and b. They rely on keeping a reference model on the server and comparing the received model with it, magnifying the effect of the benign global model, clustering the received local models on the server to generate multiple global models for each cluster. However, these approaches have multiple limitations, such as privacy issues and the possibility of making the local model unbalanced and consuming the server's resources to generate multiple global models even for the compromised clients in a cluster. In this thesis, with Weighted Federated Averaging (WFedAvg), we address the previous limitations for defense against malicious clients in federated learning for event verification. By increasing the number of benign local models and lowering the effect of the compromised clients on the server before aggregation of the model, the effects of the benign clients will increase. Furthermore, this approach has two variations, first, being that the received feedback in addition to local models on the server, will be compared with others and cosine similarity will indicate how much similarities they have, which shows the effect of different clients, second variation, which is extended of the existing works, a sample reference model will get stored on the server and in the cases where the benign clients get deviated from the sample model, they will not be considered as malicious, instead, we extend it by clustering the models and measure how far they are from the cluster that has the reference model. Then, based on the similarity of the clients in that cluster, we control the contribution level of those clients. In this way, it would be possible to get to the minimum loss value and higher accuracy much faster, as is shown in the experiment section. Additionally, while we were investigating the IoT chipset for our purpose, we discovered a vulnerability from one of the most famous IoT chipsets that can be an entry point to compromise a client for performing a federated learning attack for sensor verification, which will be discussed
CapMan: Detecting and Mitigating Linux Capability Abuses at Runtime to Secure Privileged Containers
No full textLinux capabilities represent an important security feature for enabling fine-grained management of privileges. However, limitations in selectively enabling capabilities for processes and lagging adoption from application developers often lead the operators to run containers with unnecessary privileges. Although this can potentially be addressed by modifying the application, minimizing the set of enabled capabilities, assigning capabilities to executable files, or using user-space utilities like Ptrace, those solutions typically require manual efforts, only provide partial protection, or incur significant overhead. In this thesis, we present CapMan, a solution that secures privileged containers by detecting and mitigating potential capability abuses at runtime. Our main idea is threefold. First, CapMan examines all capability requests made by system calls to ensure full protection. Second, CapMan performs the detection directly inside the Linux kernel to ensure its efficiency. Third, CapMan performs the mitigation in a transparent manner without requiring any change made to the application or container. We tackle several key challenges in realizing CapMan as follows: i) to ensure CapMan can cover every capability request, we study the Linux kernel source code to identify the kernel function used to handle such requests, and subsequently develop a kprobe-based kernel module to intercept those requests via that kernel function; ii) since user-space detection can introduce prohibitive delay, we design CapMan to perform its detection completely inside the kernel based on lightweight whitelisting and machine learning methods; iii) as Linux only allows the container itself to drop capabilities, CapMan performs its mitigation by overriding such rules in a safe manner using standard kernel functions and procedures.} Our evaluation of CapMan using real-world CVEs and capability abuses shows that it can mitigate all the tested capability abuses (most of which are missed by a state-of-the-art solution) with negligible performance overhead and resource consumption
The Power Politics of Regional Deindustrialization: The Cape Breton Development Corporation, State Ownership, and Pit Closure in Canada’s Coal Industry 1967-2001
Full text linkThe Cape Breton Development Corporation (DEVCO) was an ambitious attempt by the Canadian federal government to manage the deindustrialization of coal mining on Cape Breton Island, Nova Scotia. Created in 1967 by nationalizing the unprofitable collieries, DEVCO’s original mission was to incrementally close them, while fostering an alternative economic base in the area. The mines operated until 2001 when they finally closed. DEVCO has primarily been studied as an example of federal regional development policy, as it experimented with many projects to stimulate economic growth. However, DEVCO’s Coal Division has remained almost entirely unstudied, despite much more money, and outliving the regional development programs. Not only that, managed wind-down was quickly abandoned, and from 1973 the Coal Division expanded, a process that continued into the 1980s. In this thesis I argue that the Coal Division’s history significantly modifies our understanding of DEVCO, as regional development was only one factor in the crown corporation’s trajectory. Those other factors mostly related to coal, which the Canadian state was deeply entangled with through energy policy, labour relations, and political patronage. Furthermore, as a state-owned enterprise, DEVCO had key differences from private sector deindustrialization, as this formally politicized pit closure and made governments vulnerable to pressure from those most impacted. DEVCO was a unique response to deindustrialization, which has some enduring implications for fossil fuel infrastructures today
Vertical Mergers as a Strategic Response to Economic Uncertainty: Evidence from U.S. M&A Activity (2006–2020)
Full text linkThis thesis investigates the impact of vertical and horizontal mergers and acquisitions on investor wealth, specifically focusing on how economic uncertainty influences merger behavior and outcomes. The study follows Fan & Goyal's (2006) methodology and distinguishes between vertical and horizontal mergers using input-output tables from the U.S. Bureau of Economic Analysis. The first hypothesis posits that economic uncertainty increases the proportion of vertical mergers. In contrast, the second explores whether vertical mergers generate higher abnormal returns in uncertain times compared to unrelated mergers. Using a sample of over 100,000 U.S. M&A transactions from 2006 to 2020, sourced from SDC Platinum, the research classifies mergers by vertical relatedness and applies event study analysis and multiple linear regression methodology. Findings reveal a consistent statistical relationship between economic health indicators (S&P 500 variations) and the proportion of vertical mergers. Furthermore, results from event studies and regression models suggest that vertical relatedness has a significant positive effect on CAAR during periods of economic instability, particularly from 2006 to 2010 and 2016–2020. The Economic Policy Uncertainty Index (EPUI) support the selection of instability periods and confirms the findings between economic instability, uncertainty and CAR variations. The study contributes to the M&A literature by demonstrating that vertical integration may serve as a strategic risk-mitigation tool during volatile economic conditions, delivering higher value to investors than horizontal mergers under similar circumstances
Foreword: On Platforms—Three Approaches
Full text linkThis Foreword to the book offers three methodological approaches to platforms, which are reflected in the volume
Developing Computer Vison-based Digital Twin for Vegetation Management Near Power Distribution Networks
Full text linkThe rapid development of digital twin technology has opened new avenues for infrastructure management, particularly for addressing vegetation encroachment risks near power lines. This paper builds upon our previous work in LiDAR-based proximity detection by proposing a framework for creating digital twin for vegetation management near power distribution networks. The framework leverages the RandLA-Net model for semantic segmentation of power lines, poles, and vegetation followed by clustering and rule-based thresholding for data refinement. Detecting vegetation encroachment is achieved through KDTree-based spatial analysis, ensuring efficient identification of risk zones. The segmented and processed point cloud data is then transformed into detailed 3D models, forming the basis of the digital twin, which can be enhanced in the future by adding advanced semantic attributes and predictive tree growth models, enabling proactive vegetation management. The methodology is demonstrated through a case study, highlighting its potential to enhance operational efficiency and the resilience of power distribution networks
Art Therapy for Preverbal Trauma: Integrating the ETC for Sensory-Based Treatment
Full text linkThis theoretical intervention research explores the design of a sensory-based art therapy program for children aged four to five who have experienced preverbal trauma. Grounded in the sensory level of the Expressive Therapies Continuum (ETC), the proposed model addresses the unique therapeutic needs of young children whose traumatic experiences occurred before the development of language. Drawing on trauma theory, neurodevelopmental research, and sensory integration practices, the program offers a developmentally appropriate, trauma-informed framework that supports emotional regulation, nervous system healing, and relational repair through art-making. The methodology employs a theoretical intervention research design, with a focus on formative program development rather than implementation or outcome testing. This research contributes to the growing body of literature on early childhood trauma and highlights the potential of sensory-based art therapy as a vital avenue for healing preverbal trauma
Gender Identity Development in School-Age Children: A Cross-Cultural Examination of Identity, Emotional Well-being, and Academic Self-Concept
Full text linkThis research includes three studies that contribute to developmental psychology and gender studies by exploring gender identity development in diverse cultural contexts during middle childhood. By examining gender identity through a multidimensional lens and considering cultural and social factors, this work seeks to move beyond traditional binary views of gender. A cross-national approach, comparing samples from Montréal and Barranquilla, offers insights into both individualized and culturally specific features of gender identity development. A nuanced understanding of these relations can inform culturally sensitive and gender inclusive practices that support gender development in childhood. Study 1 investigated the replicability and generalizability of a dual-identity model of gender among children aged 10-12 years. Using cross-national samples, four gender identity clusters were identified through K-Cluster means analyses. Longitudinal data revealed changes in children's identification with gender traits over the school year, particularly those associated with the opposite gender. These changes varied across socioeconomic and demographic contexts. Study 2 examined the relations between gender identity, peer victimization, gender pressure, and anxiety among fifth and sixth-grade students. Using comparative and regression analyses, it assessed mean differences in these variables across gender clusters and how the interaction of gender identity, anxiety, and gender pressure predicted peer victimization across sociodemographic groups. The study reveals that children who identify with both gender features report the highest levels of peer victimization,
anxiety, and gender pressure. A statistically significant interaction shows that these dynamics are more pronounced in Barranquilla than in Montréal, underscoring the role of sociocultural contexts in shaping these relations. Study 3 explored the relation between peer-assessed school
performance and self-perceived cognitive competence, considering gender-related traits and contextual variables. Key findings using structural equation modelling reveal that peer-assessed competence is more strongly associated with self-perceived competence for upper-middle-class counterparts. This association is weaker for girls than boys, potentially due to SES influences on boys' academic trajectories in STEM fields. Additionally, communal traits such as being affectionate, sympathetic, understanding, and sensitive to the needs of others are more strongly associated with self-perceived competence for girls. In sum, these findings highlight the importance of recognizing diverse cultural messages that shape children's gender identities. This work can better inform educators and clinicians to support fluidity in gender expression as a normative part of development, fostering inclusive environments that promote positive identity development while also respecting diverse expressions across cultural contexts
An Effective Large Language Model-based Pipeline to Preprocess Narrative Electronic Medical Records Data for Hospital Adverse Events Detection
No full textNarrative Electronic Medical Record (EMR) data is a valuable but challenging resource for
analysis due to the need for preprocessing that comprises three essential tasks: section detection,
text normalization, and feature engineering. This thesis endeavors to establish a pipeline leveraging
Large Language Model (LLM) for the preprocessing of narrative EMR, with the objective of
identifying Hospital Adverse Event (HAE). The proposed pipeline aims to enhance the efficiency of
HAE detection by utilizing LLM, while simultaneously reducing reliance on labor-intensive, time-consuming, and costly procedures. The detection of HAE is typically accomplished through a variety of methods, which include manual chart reviews, discharge diagnostic coding, prevalence surveys, and incident reporting systems. Recently, there has been a growing interest among researchers in leveraging narrative EMR data, along with Natural Language Processing (NLP), Machine Learning (ML), and LLM techniques. A significant challenge associated with these techniques is the critical need for preprocessing narrative EMR data. Additionally, it is noteworthy that the existing tools intended for the preprocessing of narrative EMR are predominantly designed for general applications rather than being specifically optimized for the detection of HAE.
This thesis examines the preprocessing of narrative EMR to identify HAE by developing a
pipeline based on LLMs. First, given the increasing use of NLP and, consequently, LLM for HAE
detection, a systematic scoping review is conducted on this topic to summarize the existing literature, find the overlooked research gaps and the challenges related to using narrative EMR to
detect HAE. The review also underscores the essential role of preprocessing tasks in enhancing
the performance of HAE detection. The results emphasize the importance of text normalization
and establishing feature engineering in preprocessing tasks that significantly affect HAE detection
performance.
Second, the LLM-based pipeline tackles the challenges associated with the section detection
task by designing and implementing a novel multi-head attention mechanism aimed at training
LLMs for the accurate identification of section headers within clinical notes. In contrast to the
regular attention mechanisms that analyze all tokens within the input sentence, the proposed customized multi-head attention mechanism selectively directs attention towards tokens that denote
section header titles during the training phase of the LLMs. The results indicate that our approach
resulted in enhanced performance across three distinct LLMs, namely Text-to-Text Transfer Transformer
(T5), Generative Pre-trained Transformer (GPT)-2, and Bidirectional Encoder Representations
from Transformers (BERT). Notably, consistent improvements were observed in T5, which is
a smaller model.
Third, to address the text normalization challenges, a framework is proposed for detecting and
deciphering abbreviations within clinical text, employing LLMs. This framework is structured into
four distinct phases: task definition, properties identification, example selection, and the application
of LLMs through either a fine-tuning approach or an optimized example-based prompting method.
The results demonstrate that the fine-tuning approach for LLMs yields superior performance at a
lower cost compared to the optimized example-based prompt. This finding indicates that fine-tuning
LLMs effectively and efficiently facilitates the detection and deciphering of abbreviations in clinical
notes.
In conclusion, this thesis posits that customized attention directed toward the specific target task
in LLMs significantly enhances both the effectiveness and efficiency of task performance. This
customization may be achieved through various approaches, including the design of a customized
multi-head attention mechanism during training, the formulation of engineered prompts, or the systematic fine-tuning of LLMs