1722 research outputs found
Sort by
Inside out: transforming images of lab-grown plants for machine learning applications in agriculture
Get PDFMachine learning tasks often require a significant amount of training data for the resultant network to perform suitably for a given problem in any domain. In agriculture, dataset sizes are further limited by phenotypical differences between two plants of the same genotype, often as a result of different growing conditions. Synthetically-augmented datasets have shown promise in improving existing models when real data is not available."This work was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant program (Nos. RGPIN-2018-04088 and RGPIN-2020-06191), Compute Canada (now Digital Research Alliance of Canada) Resources for Research Groups competition (No. 1679), Western Economic Diversification Canada (No. 15453), and the Mitacs Accelerate Grant program (No. IT14120)."https://www.frontiersin.org/articles/10.3389/frai.2023.1200977/ful
Incarceration and Re-entry for Provincial Prisoners: Is there Hope?
Get PDFThis thesis explores the impacts of hope on the experience of incarceration and re-entry into the community and how ex-prisoners make sense of and foster hope in their lives. This research was conducted through qualitative interviews with nine participants in the Province of Saskatchewan, Canada. Qualitative interviews discerned four main themes emerging: the overall experience of provincial incarceration and the positive and negative aspects of that experience, the process of re-entry and the barriers that ex-prisoners face in that process, the motivations to change and begin desisting from crime and finally, the way in which ex-prisoners feelings of hope impacted their experiences throughout this process. Amongst the key findings is, the participants who lacked the feeling of hope during their imprisonment generally felt more disoriented during their re-entry in the community. Next, although prisons are painful and negative spaces, they can also offer an opportunity for reinvention (Crewe and Ievins, 2020). As such, many of the participants in this research found various ways to “reinvent” themselves post-imprisonment. In alignment with procedural justice literature, this research found that when prisoners had positive interactions with correctional officers, it profoundly impacted their feelings of hope. As such, the participants who had procedurally just interactions described feeling more hopeful in their prison sentence and later re-entry into the community. This research also found that although these ex- prisoners had shorter sentences, they still felt various strains in their process of re-entry. The most predominant way that the participants navigated their re-entry into the community and further desistance from crime was by distancing or “knifing off” old peers (Laub & Sampson, 2001). Despite having difficulty interpreting hope, after thought and reflection, each participant made sense of the term by attributing it to a more positive future.Master of Arts in Criminal Justic
From Posthumanist Anaesthetics to Promethean Dialectics: Further Considerations on the Category of the Hysterical Sublime
Get PDFThis essay proposes a critique of posthumanist critical theory through the development of the category of the hysterical sublime, a concept first introduced by Fredric Jameson in his early writings on postmodernism. While some critical posthumanist theories equate representationalism with a transcendental humanism, representation is inherent to the kinds of abstractions required in theory as such. Taking up the posthumanist resistance to anthropocentrism or human exceptionalism, the concept of the hysterical sublime—as opposed to the category of the modern sublime, regarding nature—is used to convey the way that posthumanism undermines its own ethical injunctions by invoking fatalistic representations of human action. Instead, this essay defends the return to dialectical humanism as an appropriate framework for thinking the rational and material conditions required for ethical human action.SSHRC 430-2020-00738https://www.tandfonline.com/doi/full/10.1080/08935696.2023.218368
Securing Intrusion Detection Systems in IoT Networks Against Adversarial Learning: A Moving Target Defense Approach based on Reinforcement Learning
Get PDFInvestigating the use of moving target defense (MTD) mechanisms in IoT networks is ongoing research, with unfathomable potential to equip IoT devices and networks with the ability to fend off cyber attacks despite the computational deficiencies many IoT ecosystems typically have. The AI community has extensively studied adversarial threats and attacks on machine learning-based systems, emphasizing the need to address the potential compromise of anomaly-based intrusion detection systems (IDS) through adversarial attacks. Another concept that has gained significant attention in the networking community is Game Theory. Protecting any given network is almost a never-ending battle between the attacker and defender, and hence a natural game of competitors can be modelled based on one’s parametric specifications to gain more insight into how attackers might interact with one’s system. The goal of this thesis is to propose a comprehensive, experimentally verifiable game-theoretic model of MTD in IoT networks to secure the IDS against adversarial attacks. Once a game with state transitions based on given actions can be modelled, reinforcement learning is used to develop policies based on various episodes (rounds) of the game, ultimately optimizing network decisions to minimize successful attacks on machine learning-based IDS. The state-of-the-art ToN-IoT dataset was investigated for MTD feasibility to implement the feature-based MTD approach. The overall performance of the proposed MTD-based IDS was compared to a conventional IDS by analyzing the accuracy curve of the MTD-based IDS and the conventional IDS for varying attacker success rates and resource demands. Our approach has proven effective in securing the IDS against adversarial learning.Master of Science in Applied Computer Scienc
Dynamic perspectives on education during the COVID-19 pandemic and implications for teacher well-being
Get PDFTwenty teachers took part in bi-weekly interviews over the course of the 2020–2021 school year and again one year later during the COVID-19 pandemic. Comparative findings on teachers’ experiences indicated varied circumstances and a wide array of perspectives on coping during this protracted and stressful time. While some teachers demonstrated flourishing and resilience, the majority experienced a tipping point toward burnout. A small group languished, relating indicators of burnout and post-traumatic stress. Given the dynamic findings, a continuum of awareness is suggested that might assist teachers and administrators in critically assessing the range and dimensions of coping exhibited during the pandemic or subsequent stressful periods of time. With information of this nature available, we propose that school organizations could be better informed to provide supports and resources and improve worklife balance and well-being of teachers."We would like to acknowledge the Social Sciences and Humanities Research Council of Canada (SSHRC) that enabled us to conduct this research. We thank SSHRC for the support in the form of Partnership Engage Grant #1008-2020-0015."https://www.sciencedirect.com/science/article/pii/S266637402300016
Source and fate of dissolved organic matter in boreal headwater streams
Get PDFUnderstanding the source and fate of dissolved organic matter (DOM), a key water quality variable, in boreal headwaters is of critical importance considering the amount of carbon stored and processed in different ecosystem components within the boreal forest and the sensitivity of these processes to climate change. Using historical streamflow and stream chemistry data in combination with direct measurements of the landscape sources of DOM and more detailed stream DOM quality data from 2021 at the IISD-ELA, I examined how the terrestrial source of DOM influences the quantity and quality of DOM in three boreal headwater streams. Using historical stream data from 1981-2021, I found that concentration-discharge (c-Q) relationships varied based on both catchment characteristics and hydrological conditions. Streams draining upland-dominated catchments were more often transport-limited (i.e., concentration increased with increasing flow), whereas a wetland-dominated stream was more often source-limited (i.e., concentration decreased with increasing flow) in terms of stream DOM concentration. DOM concentration and quality data in soil leachate indicated that streamwater had DOM characteristics suggesting it originated from near-stream organic soils, while after the drought the DOM came proportionally more from distal mineral soils (in addition to near-stream organic soil contributions). I showed that the severe drought in 2021 made streams with varying landscape characteristics respond similarly to the post-drought flush. These findings also illustrate that while c-Q relationships may be different among streams draining upland-dominated and wetland-dominated catchments as a result of the different abilities of these landscape to accumulate and mobilize DOM, DOM quality responded to this drought to post-drought flush synchronously among all three streams. As climate change will alter the frequency, duration, and severity of future hydrological conditions, this has repercussions for the DOM dynamics in headwater streams and the resulting water quality downstream."I was supported in this research by a UWGSS Scholarship from UWinnipeg and an NSERc - Canada Graduate Scholarship."Master of Science in Bioscience, Technology and Public Polic
An Efficient Approach to Compute Zernike Moments with GPU-Accelerated Algorithm
Get PDFThe utilization of Zernike moments has been extensive in various fields, including image processing and pattern recognition, owing to their desirable characteristics. However, the application of Zernike moments is hindered by two significant obstacles: computational efficiency and accuracy. These issues become particularly noticeable when computing high-order moments. This study presents a novel GPU-based method for efficiently computing Zernike moments by leveraging the computational power of the Single Instruction Multiple Data(SIMD) architecture. The experimental results demonstrate that the proposed method can compute Zernike moments up to order 500 within 0.5 seconds for an image of size 512 * 512. To achieve greater accuracy in Zernike moments computation, a k * k sub-region scheme was incorporated into the approach. The results show that the PSNR value of the Lena image reconstructed from 500-order Zernike moments computed using the 9 * 9 scheme can reach 39.20 dB. Furthermore, a method for leaf recognition that leverages Zernike moments as image features, with the k-Nearest Neighbors (k-NN) algorithm serving as the classifier is proposed. The proposed method is evaluated on the Flavia leaf dataset, and the results affirm the effectiveness of the approach.Master of Science in Applied Computer Scienc
A Survivor’s Narrative of Institutional Harms Experienced in Manitoba Developmental Centre and Prisons in Canada
Get PDFThis thesis is a case study that looks at the life story of a Black man, Dwight, who had been institutionalized at the Manitoba Developmental Centre (MDC) at the age of eleven in 1967 for three years. Dwight alleges that he experienced physical and sexual abuse, which led him to using violence as self-defence against older patients until he was expelled from the institution. Shortly thereafter, Dwight entered the correctional institutions in Canada, being incarcerated in sixteen prisons after the MDC. This thesis applies Goffman’s (1961) theory of “total institutions” and critical disability theories. These theories are used to contextualize Dwight’s perspective of his experiences on the processes of institutionalization, roles in institutions, and the social construction of disabilities. Oral history or life story interviews were conducted with Dwight to gain insight regarding his institutional experiences. These life story interviews cover his life prior to institutionalization at the MDC, his time inside, and his life in the community. There are several research contributions and implications. This case study provides an outline of the Black experiences in institutions for persons with disabilities. Other research contributions are that misdiagnosis does occur and in this case was socially constructed based on the time period of the 1960s. This study contributes to research that shows that race may affect the experiences of labelling as current studies reveal that Black youth continue to be more likely to be assigned a disability diagnosis when compared to their white peers. Finally, the importance of oral history as a methodology provides rich detail and new knowledge from lived experiences that other methods may not provide in the fields of sociology, criminology, and history. Future research with survivors of MDC and other institutions may benefit using oral history as it is flexible and allows for participants to share as much as they wish. This method of inquiry also allows interviewees to be heard as persons with lived experiences of disabilities have experienced marginalization in the historical narrative of institutions.SSHRCMaster of Arts in Criminal Justic