21793 research outputs found
Sort by
Fall Prevention for Older Women Using Online Dance Classes with Blood Flow Restriction
Get PDFThis project examined if online dance classes could provide a safe and accessible method for older women to improve their physical activity levels and reduce their risk of falls. Women aged 65 years and above were recruited to complete 12 weeks of twice-weekly 75-minute interventions, and were evaluated pre, mid and post via 30-second trials of quiet standing, the star excursion balance test (SEBT), 30-second Sit-to-Stand (30-CST) and the Calf Raise Senior (CRS). Significance was evaluated using non-parametric statistics (p≤.05). Participants demonstrated high attendance rates (80.4 ± 13.8%), decreased mediolateral sway during eyes closed (pre-mid p=.003) and foam conditions (pre-mid p=.02), with smaller sway area for foam conditions (pre-mid p=.015), larger reaches on the SEBT (lateral: pre-mid p=.008, pre-post p=.008; posterior-lateral: pre-post p=.009) and higher number or CRS repetitions (mid-post p=.02, pre-post p=.015). A follow-up study was conducted to try and overcome intensity limitations encountered with the online environment by using blood flow restriction (BFR). Participants completed 12-weeks of online dance classes with half the group randomized to wear BFR cuffs. No improvements were found among the control group. Participants in the BFR group demonstrated increases in strength on the 30-CST (pre-mid p=.042; pre-post p=.039) and greater reaches on the SEBT in medial (mid-post p=.043) and posterior-medial directions (pre-post p=.043). Online dance classes are an effective, safe and accessible fall prevention program and the addition of low-cost BFR cuffs further enhances strength and dynamic balance, thereby increasing independence and quality of life through older age
Computation-Efficient CNN System for High-Quality Lung Nodule Detection
Get PDFLung cancer diagnosis is a critical healthcare issue, and fully automated lung nodule detection is desirable for a timely diagnosis. However, due to the variability in shapes, sizes, textures, and locations in lung nodules, developing a computer vision system for this detection is a very challenging task.
In this thesis, a special CNN system is proposed for lung nodule detection. It consists of 2 stages, namely Stage A and Stage B. Stage A is designed to localize the nodule candidates, aiming at a high sensitivity in order to minimise the miss rate. Stage B is to identify the true nodules from the input samples. It can be used to identify falsely detected nodule samples from the output of Stage A, and also as a stand-alone lung nodule recognition system.
In Stage A, there are three blocks, i.e., a pre-processing block, custom-design CNN block and refinement block. The core of this stage is the custom-designed and U-net-based CNN block. The filtering modules in its convolution layers are specifically designed to suit the features produced in these layers. To reduce the data loss in the first four layers, Full-ReLU is used as the activation function. Furthermore, the refinement block is placed to reduce effectively the false positive rate. The computation complexity of Stage A is very low, as its total number of trainable parameters is only 0.16 M. Stage A delivers a high detection rate of 95.38% but the false positive rate is still as high as 6.9 FPs/scan. The output data will be applied to Stage B for further processing.
The design of Stage B is focused on distinguishing between the true nodules and their look-likes. Based on our analysis on the characters carried by nodules of different sizes, we propose to have 2 networks in Stage B for large and small nodule categories, respectively. The feature extraction in the 2 CNNs should be different, one targeting the variations in object regions of large nodules and the other looking more into nodule surroundings in case of small nodules. Two CNNs have been designed and each of them has a particular multi-branch feature extraction (FE) block for the designated nodule category. Each CNN also involves fully-connected layers for classification. Stage B has been tested as a stand-alone lung nodule recognition system on LUNA 16 dataset. The results demonstrate that, with respect to similar systems found recently in literature, Stage B provides a good processing quality at a computation cost that is only a very small fraction of that needed by others.
The complete system for lung nodule detection, i.e., Stage A and Stage B combined, has also been tested with the same dataset. The results demonstrate the good functionality of the system. All these CNNs combined require 0.7M parameters, far less that other CNN systems performing the same task.
In summary, the proposed system has been custom-designed to optimize the computation efficiency, i.e., achieving a good detection quality at the lowest computation cost. To attain this goal, the design strategy is to decompose the complex task of lung nodule detection into subtasks so that the system can employs multiple simple CNNs, each performing a sub-task. In this way, each CNN can be structured to suit the characters of a particular kind of nodule data and optimised to meet specific performance requirements. The effectiveness of this strategy has been confirmed by the results of the performance evaluation. Because of its low computation cost, the proposed system can be very easily implemented in various environment
Graph Representation Learning for 3D Human Pose Estimation
Get PDFGraph convolutional networks (GCNs) have proven to be an effective approach for 3D human pose estimation. By naturally modeling the skeleton structure of the human body as a graph, GCNs are able to capture the spatial relationships between joints and learn an efficient representation of the underlying pose. However, most GCN-based methods use a shared weight matrix, making it challenging to accurately capture the different and complex relationships between joints. In this thesis, we introduce an iterative graph filtering framework for 3D human pose estimation, which aims to predict the 3D joint positions given a set of 2D joint locations in images. Our approach builds upon the idea of iteratively solving graph filtering with Laplacian regularization via the Gauss-Seidel iterative method. Motivated by this iterative solution, we design a Gauss-Seidel network architecture, which makes use of weight and adjacency modulation, skip connection, and a pure convolutional block with layer normalization. Adjacency modulation facilitates the learning of edges that go beyond the inherent connections of body joints, resulting in an adjusted graph structure that reflects the human skeleton, while skip connections help maintain crucial information from the input layer’s initial features as the network depth increases. Our experimental results demonstrate that our approach outperforms the baseline methods on standard benchmark datasets.
This thesis makes another significant contribution by designing a spatio-temporal 3D human pose estimation model. Accurate 3D human pose estimation is a challenging task due to occlusion and depth ambiguity. To address these issues, we introduce a novel approach called Multi-hop Graph Transformer Network, which combines the strengths of multi-head self-attention and multi-hop graph convolutional networks with disentangled neighborhoods to capture spatio-temporal dependencies and handle long-range interactions. The proposed network architecture consists of two main blocks: a graph attention block composed of stacked layers of multi-head self-attention and graph convolution with learnable adjacency matrix, and a multi-hop graph convolutional block comprised of multi-hop convolutional and dilated convolutional layers. Extensive experiments demonstrate the effectiveness and generalization ability of our model, achieving state-of-the-art performance on benchmark datasets while maintaining a compact model size
Modelling reindeer rut activity using on-animal acoustic recorders and machine learning
Get PDFResearchers have been using sound to study the biology of wildlife to understand their ecology and behaviour for decades. By gathering audio from free-ranging species using on-animal recorders, their vocalizations can be used to describe their behaviour and ecology through signal processing. Unfortunately, processing hours of recordings is incredibly time-consuming. By applying machine learning to audio recordings, researchers have used neural networks to decrease the processing time of acoustic data. However, until now, most of this research has focused on analyzing the data of stationary recorders. To show the utility of on-animal recorders in combination with machine learning, we recorded the vocalizations of reindeer (Rangifer tarandus) during their rut at the Kutuharju research station in Kaamanen, Finland. We used vocalizations as an activity index to describe the rut activity of male reindeer. In 2019 and 2020, we placed recorders around the necks of seven reindeer during their rut. We trained convolutional neural networks to identify reindeer grunts, which were then used to classify their vocalizations. Of the networks’ vocalization classifications, around 95% of them were correct. With such high metrics, we could reliably explore the males' activity patterns using a neural network. We then analyzed the reindeers’ vocalization using generalized additive models. The patterns suggested heavier, older males vocalized more than lighter, younger males and, overall, were more active during the day than night. Overall, on-animal acoustic recorders, in tandem with machine learning, proved to be effective tools, and with more attention, they could prove valuable tools for other researchers
A Fluorescence-Based Coupled Enzyme Cascade Assay in the Investigation of Old Yellow Enzymes for Biopolymer Production
Get PDFPlastics are a useful and necessary material in the modern world. However, the methods of extraction and the finite nature of petroleum necessitates divesting from traditional petroleum-derived plastics. One avenue being pursued is sustainably sourced plastics; namely plastics made from a biorenewable starting source. Enzyme catalysis and Old Yellow Enzymes (OYEs) in particular are of note in pursuing the generation of plastics from biorenewable sources. Engineering OYEs in pursuit of expanded substrate scope, improved efficiency, and other traits is growing increasingly popular and tools towards this aim remain valuable. In this work, I characterized, optimized, and utilized a novel fluorescence-based enzyme cascade in the investigation of nine candidate OYEs and their activities on four biorenewable plastic precursors. This enzyme cascade couples the redox activity of OYEs to the release of 4-methylumbelliferone, a fluorophore with a fluorescence intensity maximum at 445 nm. First the cascade was verified using Malate Dehydrogenase. This was followed by the optimization of the enzymatic cascade through the mutation of supporting enzyme GapA to utilize NADPH instead of its native NADH and the adjustment of NADPH concentrations to limit off-target fluorescence. Once optimized, the assay was utilized in a screen of nine OYEs with four biorenewable plastic precursors. Hits from the screen were investigated and the assay itself was investigated in the pursuit of enzymatic characterization. With a successful use-case demonstrated for this robust fluorescence-based activity assay, I was successfully able to add a tool to the development of OYEs towards generating sustainable plastics
An Experience That Lasts a Lifetime: Building Modernity, Man, and Nation at the YMCA of Montreal's Kamp Kanawana, 1894-1967
Get PDFWith their stunning lakefront views and pristine forest trails, summer camps have been a hallmark of a Canadian summer since their inception in the 1890s. Established out of anti-modernist and anti-urbanist sentiments among the upper-middle class, summer camps have been cemented as an important rite of passage for Canadian youth. They are also integral in the construction of national and masculine identities. This thesis takes one of the oldest summer camps in Canada and considers how the overlapping currents of religion, colonialism, and national identity practice took shape at the YMCA of Montreal’s Kamp Kanawana. Located an hour north of Montreal, Kanawana has operated since the 1890s and thousands of young children have passed through its gates. From 1894 to 1967, Kanawana was open only to boys and through its varied programming, helped form distinctly masculine and Canadian young men. This thesis is divided in three thematic chapters that each look at an important phase of Kanawana programming: active Christian citizenship, playing Indian, and the myth of the voyageur. Drawing from archival research and personal experiences at a different YMCA camp, this thesis explores the legacy of summer camps like Kanawana in the Canadian imagination
Beyond the walls of classrooms: Exploring the pedagogical effectiveness of text-to-speech-based shadowing on the development of Mandarin tones
Get PDFWith limited classroom time (Collins & Muñoz, 2016), teachers struggle to provide personalized language input (listening activities) and opportunities for students to practice output (speaking). Text-to-speech synthesizers (TTS), also known as text readers, offer a possible solution by allowing students to interact with the computer anytime-anywhere, and at their own pace (Cardoso, 2022). As such, this technology has the potential to improve students' listening skills and provide flexible language practice (Little, 1995). Although TTS offers many benefits (e.g., immediate access to the language; Liakin et al., 2017), an unresolved issue is that the technology does not incorporate an output-inducing component (Fang, 2017). To address this issue and contribute to the field of computer-assisted pronunciation instruction, this study combines TTS with shadowing (i.e., the repetition of a word or phrases immediately after hearing it; Lambert, 1994), a technique that has been proven to be effective in developing L2 pronunciation (Foote & McDonough, 2017; Zajdler, 2020). By combining these two technologies, to which we will refer as “TTS-based shadowing training” (TTS-S henceforth), our approach provides learners with the benefits of both TTS (exposure to input) and shadowing (opportunities to practice output).
To determine the probability of success of this innovative approach, this study examined the pedagogical effectiveness of using TTS-S in a self-regulated learning environment to acquire tones #1 and #4 in Mandarin Chinese. While tone #1 (high tone) is relatively easy to acquire in comparison with other tones, tone #4 (descending tone) is considered one of the hardest to produce (Hendry, 2023). The research was guided by the following research question: can TTS-S help L2 learners raise their sound awareness and improve their perception and production of the target Mandarin tones over six weeks? By means of pre-/post-tests (to assess effectiveness in pronunciation), ten beginner-level participants were asked to complete: (1) an awareness task in which they verbalize their metacognitive knowledge of Mandarin tones; (2) ABX tasks to assess their perception of Mandarin tones; and (3) a production task to evaluate the production of the target tones. Results indicate that the use of TTS-S did not yield significant enhancements in terms of awareness, perception, and production, possibly due to the presence of a ceiling effect in some of the measures adopted and other methodological limitations
ADDICTION > recovery: The Surprising Spiritual Solution to Masculinity and Deindustrialization Happening in a Men’s Addiction Treatment Centre
Get PDFRecovery from addiction is often viewed as the achievement of abstinence as well as the reclamation of the things lost to addiction: agency, health, productivity, spiritual wellness. This study, based on fieldwork at a Twelve Step-oriented men’s treatment centre in Vancouver, recasts addiction and recovery as a problem and solution on opposing sides of an unbalanced equation, where the problem is greater than the solution: ADDICTION > recovery. With the solution retroactively oriented to the individual’s unwellness, the social conditions that continue to produce addiction remain. It is argued that the treatment centre in this study is “treating” three intersecting “crises,” not just one: the drug and addiction crisis, the “masculinity” crisis, and deindustrialization. As such, attention is given to the conditions affecting the overrepresented treatment residents encountered in this study, known to be dying in the toxic drug “epidemic” at rates disproportionate to other groups (Perrin 2020): working class white men. These conditions include: unemployment, abandoned communities, loss of social status and security, and masculine norms that valourize risk, invulnerability, and “working hard and playing harder.” Additionally, this imbalance potentiates the exploitation of those who work and volunteer in the treatment industry, many of whom are “recovering addicts” deeply concerned with saving their “brothers.”
Still, while the recovery solution is misaligned with these problems, the spiritual, disciplinary treatment process – which encourages men to depend on God, to open up, to ask for help, and to be “of service” – does offer men in treatment a kind of moral reskilling that can mitigate the risk of working class white men turning to the dangerous politics of aggrievement, and can help orient them to the “soft skills” of the “service economy.” The recovering men who end up working in treatment can be seen as the once “hyper masculine” becoming effective workers in the historically feminized realm of carework. As such, a man who starts out injured by deindustrialization and then comes to work as a recovery/care worker is a living embodiment of the transformation of labour, of heavy labour being reborn as service work
Solution-Acceleration Strategies for High-Order Unstructured Methods
Get PDFThe design of next-generation aircraft relies on computational fluid dynamics (CFD) to minimize testing requirements at reduced cost and risk. However, current industry reliance on Reynolds-averaged Navier-Stokes (RANS)-based CFD is limited in predicting transitional and turbulent flows. Large-eddy simulation (LES) offers accuracy where RANS methods fail, but can have prohibitive computational cost. To address this, we propose a high-order CFD framework to advance flux reconstruction (FR) methods toward industrial-scale simulations. FR is a family of high-order, unstructured schemes that provide accuracy at reduced cost per degree-of-freedom (DOF) compared to low-order methods, with proven potential for LES. We develop practical strategies to reduce the computational cost of FR methods for explicit and implicit formulations. Due to the low cost per time step, explicit time stepping is typically used in FR methods. However, stability constraints prohibitively limit time-step sizes in numerically stiff problems. Hence, implicit time stepping is preferred in these cases, but it requires solving large, nonlinear systems and can be computationally expensive.
This thesis introduces optimal Runge-Kutta methods to alleviate stability limits and reduce wall-clock times by approximately half in moderately low stiffness problems. For increased stiffness, we hybridize implicit FR methods using a trace variable, which allows a reduction of the implicit system via static condensation, decreasing implicit time stepping costs, especially at higher orders. Hybridization with both discontinuous (HFR) and continuous function spaces (EFR) is suitable for advection and advection-diffusion type problems within the FR method and enables significant speedup gains over standard FR. We incorporate polynomial adaptation to the hybridized framework, varying the solution polynomial’s degree locally within each element, which results in an overall reduction in DOF and significant speedup gains in a two-dimensional problem against standard polynomial-adaptive formulations. Finally, we combine implicit-explicit (IMEX) time stepping with hybridization to tackle geometry-induced numerical stiffness. The resulting method reduces computational cost at least fifteen times over explicit methods in a multi-element airfoil problem at Reynolds 1.7 million. Our proposed framework enables substantial reductions in both moderate and high stiffness problems, thus advancing high-order methods toward large industrial-scale problems
Quebec Feminist Film Culture in the 21st Century: A Transnational Perspective
No full textThis dissertation looks at the emergence of transnational feminist audio-visual practices in Quebec that endorse, develop, and promote approaches to gender-specific issues in the film industry and cinematic imaginaries within the geo-cultural and geo-political francophone space. Drawing on transnational feminism, on media ecology’s environmental approach to media, and transnational approaches to film studies, each chapter examines the circumstances, networks, and infrastructures that enabled feminist film culture in Quebec within the larger context of global, and specifically francophone, media. In this study, I bring together disparate formations of feminist media cultures, activism in the film industry, and filmmaking practices, including online distribution and curation of video documentary and video-art, questions of female authorship, feature fiction films shown in A-list international film festivals, and diasporic filmmaking practices from postcolonial francophone subjects migrated to Quebec. My case studies demonstrate how these contemporary feminist media and filmmaking practices in Quebec are both situated within the genealogy of Quebec women’s cinema from the late 1960s and 1970s, and also transcend their nationalist theoretical and political frameworks.
Chapter 1 explores the Montreal-based, feminist/queer artist-run centre Groupe Intervention Vidéo (GIV), and its use of media technology. I draw both on feminist approaches to media technology to illustrate GIV’s use of video technologies since the 1970s as a way of doing feminist political work, and on feminist critiques of platform studies to assess how GIV adapts its working methods to streaming platforms. Chapter 2 examines contemporary configurations of Quebec women’s cinema within international film festival circuits through the work of three filmmakers, Chloé Robichaud, Sophie Deraspes, and Geneviève Dulude-De Celles. I combine the concepts of cinéma-monde and feminist theorizations of women’s cinema and female authorship to address Quebec transnational cinema from a gender-specific perspective. In Chapter 3, I critique hegemonic feminist film discourse in Quebec that universalized the category of “woman” and aligned with the dominant narrative of national identity in Quebec, which denied its colonial past and essentialized Quebec society as white and francophone. Against this backdrop, I illustrate how the work of three diasporic francophone filmmakers in Quebec - Gentille M. Assih, Hejer Charf, and Maryanne Zéhil expands dominant understanding of women’s cinema by foregrounding the gendered experience of immigrant subjects across various cinematic forms, including documentary, experimental films, and feature fiction