21793 research outputs found
Sort by
Dividend Yield as a Predictor of Stock Returns During Market Volatility: A Comparative Study of Volatile and Non-Volatile Periods in the Indian Equity Market
No full textThis study investigates the role of dividend yield as a predictor of equity premia in the Indian equity market across periods of market volatility and stability. It contributes to the broader understanding of the challenges associated with using dividend yield as a forecasting tool in an emerging market contextand highlights the limitations of relying solely onfundamental indicators. Using a simplified Fama-French framework, the research analyzes a subset of NIFTY 50 stocks that were part of the index before 2013, selected based on their longevity and consistent dividend payment history, encompassing three volatile and three non-volatile phases from 2008 to 2022. The findings reveal that dividend yield exhibits limited and statistically insignificant predictive power across all periods. Out-of-sample analyses further confirm the model’s poor forecasting performance, with large divergences between actual and predicted equity premia and negative Prediction R-squared values. Additional robustness checks were performed by incorporating sectoral indices and firm size. While sectoral effects only marginally improved explanatory power, the inclusion of firm size significantly increased R-squared in certain phases, particularly during COVID and Post-COVID, though dividend yield itself remained insignificant. The study acknowledges its limitations and offers directions for future research, including the incorporation of macroeconomic variables, the examination of sector-specific dynamics, and the use of advanced predictive modeling techniques to improve return forecasts in emerging markets
Securing Federated Learning: A Comprehensive Defence Against Privacy Attacks
No full textIn this information age, machine learning (ML) applications drive smart living through innovations such as personalized healthcare, intelligent transportation, and smart homes. However, despite these advancements, businesses and industries continue to face significant challenges in safeguarding data privacy, as ML systems increasingly rely on vast amounts of user data. In this direction, Federated Learning (FL) has emerged as a promising solution, enabling collaborative model training while keeping user data on local premises without the need to share raw data. However, FL faces significant challenges also, including expensive communication costs, system heterogeneity, and vulnerability to various attacks. In particular, it is susceptible to poisoning attacks, where malicious participants corrupt models and data as well as inference attacks that exploit gradients to reveal sensitive information through membership inference or model inversion techniques. These attacks can extract sensitive information from shared gradients, undermining the fundamental privacy guarantees of federated systems. Thus, effective defence mechanisms are fundamental for fully leveraging the advantages of FL. Numerous defences, such as FoolsGold, Flod, Flad, MADDPG, and others,
are in place to secure the FL systems.
However, the majority of the defence mechanisms suffer from accuracy degradation, computational overhead, and inadequate attack prevention. Most client selection methods cannot reliably separate malicious and straggler clients, with even cutting-edge approaches struggling with herding and cold-start issues. Furthermore, recent state-of-the-art techniques frequently fail to defend against inference attacks adequately. These methods typically employ Secure Multiparty Computation (SMPC), Homomorphic Encryption (HE), or Differential Privacy (DP) as defensive measures. However, SMPC and HE suffer from high computational complexity, while DP often leads to degraded model accuracy.
Thus, this research addresses these limitations by proposing five different defence mechanisms that ensure robust FL with protected gradients. The proposed mechanisms consist of FedChallenger, Fed-Reputed, SignDefence, Ada-Sign, and SignMPC. The proposed FedChallenger introduces a dual-layer defence mechanism that comprises the zero-trust challenge-response-based authentication at the first layer and a variant of Trimmed-Mean aggregation at the second layer that leverages pairwise cosine similarity and Median Absolute Deviation (MAD). Extensive evaluation on MNIST, FMNIST, EMNIST, and CIFAR-10 datasets demonstrates 3-10% accuracy improvement over state-of-the-art approaches with 1.1-2.2 times faster convergence and 2-3% higher F1-scores.
Subsequently, the reputation-based client selection approach, Fed-Reputed, leverages device capability information and a modified Bellman equation within a hierarchical framework, integrated into a Deep Q-Learning Network (DQN)-based Imbalanced Classification Markov Decision Process (ICMDP) classifier for enhanced client selection. Testing on MNIST and FMNIST datasets demonstrates 9-50% accuracy gains and 1.3-1.7 times faster convergence while effectively detecting both malicious and straggler clients.
Moreover, existing methods often suffer from the dying ReLU problem, where neurons permanently deactivate during training. To counter the dying ReLU problem, SignDefence implements a sophisticated aggregation scheme that utilizes sign direction and LeakyReLU-based aggregation, incorporating Jaccard similarity derived from binary-encoded model weights. This technique demonstrates consistent accuracy and F1-score improvements across different attack conditions.
Despite its benefits, SignDefence remains vulnerable to inference attacks and suffers from limited generalization due to its fixed threshold across diverse benchmark datasets. To address these limitations, a lightweight strategy, Ada-Sign, employs adaptive threshold computation and incorporates DP mechanisms. This approach maintains comparable accuracy to SignDefence while providing enhanced gradient protection through adaptive DP. Extensive evaluation on MNIST and HAR datasets reveals 3-20% accuracy improvement for Ada-Sign over the majority of state-of-the-art techniques.
Finally, to enhance the protection of both SignDefence and Ada-Sign against inference attacks, iv SignMPC integrates highly configurable SMPC, HE, and DP algorithms. This combined approach ensures comprehensive communication security and gradient privacy while avoiding significant performance bottlenecks. Comprehensive evaluation on MNIST and HAR datasets demonstrates 4-17% accuracy gains for SignMPC over established approaches while maintaining computational efficiency and robust privacy guarantees
AfroGoth: The Horrors of Anti-Black Racism in Toni Morrison’s Beloved and Jordan Peele’s Get Out
No full textThough the gothic has historically been viewed as European and an implicitly white genre, it is now a powerful tool for exposing anti-Black racism. This thesis argues that Toni Morrison’s novel Beloved and Jordan Peele’s film Get Out use gothic scenes of Black suffering in order to illustrate and confront the horrifying effects of white supremacy on Black bodies. We will emphasize that the Gothicism of Beloved and Get Out exposes and criticizes “horrifying whiteness” not only by inverting Eurocentric gothic traditions but also by drawing on traditional African figures like the watermeisie and the zombi, and the spiritual values associated with them. Finally, we will show that because of the structural violence of supposedly helpful mainstream white institutions of healing and care, both works continue to be extremely relevant
“I take too much time screaming at articles and just critiquing”: An Examination of the News
No full textThis thesis research project examines the quality of news reporting on and about people who use drugs in Newfoundland and Labrador. This is done by examining news reporting and through interviews with those working in the harm reduction field. A reflexive thematic analysis (RTA) is conducted on 55 news articles from the five major news outlets in Newfoundland and Labrador to illuminate how local journalists reported on the “Towards Recovery” reports, an initiative created by the government to assist in overhauling the province’s mental health and addictions services. The RTA of the news articles found that journalists consistently failed to source from people with lived experience of drug use, instead prioritizing government and healthcare officials, while failing to provide actionable information to the reader. Following the article RTA, this thesis research project also employed semi-structured interviews with five harm reductionists in Newfoundland and Labrador’s capital city of St. John’s, asking how they thought local journalism can better support people who use drugs and harm reductionists. The interviews revealed a desire for more collaboration between harm reductionists and journalists. This research’s combined results reveal a gap that a collaborative form of journalism, like solutions journalism, may be equipped to fill. This study has practical implications for local journalism in Newfoundland and Labrador since it shows a clear need for a paradigm shift: Journalism must view people who use drugs as part of the audience, not as extensions of their audience
Comparative Study of Electrochemical Pretreatment of Sludge for Enhancing Green Energy Production
No full textAnaerobic digestion (AD) reduces greenhouse gas emissions by converting sludge into renewable energy such as biogas and hydrogen. However, the compact structure of raw sludge limits biodegradability and biogas yield. Pretreatment can enhance hydrolysis and improve energy recovery. This study investigates electrochemical (EC), thermal-pressure (TP), and combined TP-EC pretreatments and their effects on sludge properties and biogas production. The research was conducted in three phases. In Phase 1, EC pretreatment was tested using three reactor configurations. The reactor with 213 mm electrode spacing outperformed the 50 mm reactor, achieving 89% phase separation after 300 minutes at 1.3 V/cm. Scaling up with an enhanced 3.4 V/cm voltage gradient for 60 minutes significantly increased sludge temperature, solids content, and organic compound solubilization. The SCOD/TCOD ratio in EC-pretreated sludge was 11 times higher than in raw sludge, demonstrating scalability. Phase 2 examined TP pretreatment (90–180 °C, 0.6–1.3 MPa) and TP-EC combinations. TP pretreatment enhanced solubilization but consumed more energy than EC. For example, TP pretreatment at 165 °C and 1.1 MPa for 30 minutes yielded an SCOD/TCOD ratio 7 times higher than raw sludge but 1.5 times lower than EC. TP-EC sequences improved solubilization but required even higher energy input. Phase 3 compared all methods through physicochemical analyses, BMP test, LC-MS, and SEM. EC pretreatment achieved the highest biogas yield: 83% more than raw sludge, 42% more than TP-pretreated sludge, and 11% more than TP-EC sludge. These findings demonstrate the potential of EC pretreatment to enhance biogas production and support its integration into existing and new anaerobic digesters, including those in small and medium-sized municipalities
Enhanced Video Tracking Based on Fusion of Visible and Infrared Images
No full textVideo tracking is the process of automated identification, localization, and continuous monitoring of objects of interest throughout consecutive video frames. Video tracking is core of many cutting-edge vision applications such as surveillance systems, autonomous vehicles, augmented reality, robotics, and human-computer interaction. However, reliance solely on visible (RGB) imagery introduces significant challenges, including poor visibility, low illumination, occlusion, and appearance variations. To overcome these challenges, fusion of RGB with thermal infrared (TIR) data has been explored to leverage complementary modality information for improved tracking performance under challenging conditions.
Many existing RGB-Thermal (RGB-T) trackers use deep learning (DL) methods for strong object feature representation. However, despite their superior performance, these tracking methods mostly rely on dual-branch architectures, complex fusion modules, or external teacher-student frameworks, leading to increased model size and significant training overhead. To address this problem, this thesis proposes unified RGB-T tracking schemes that enhance conventional RGB trackers without altering their network architectures or significantly increasing the computational complexity.
In the first part of the thesis, a novel pixel-level fusion network, symmetric bidirectional dynamic fusion (SBiDF), is introduced. SBiDF enhances RGB inputs by dynamically integrating TIR data at the pixel level prior to tracking, utilizing modality-specific autoencoders, dynamic convolutional filtering (DCF) blocks, and an output fusion module. The DCF blocks perform adaptive, bidirectional, content-aware enhancement, enabling balanced cross-modal refinement. Importantly, SBiDF generalizes effectively beyond TIR to additional modalities such as depth and event data, providing superior tracking accuracy and broad applicability without modifying the tracker architecture.
The second part of the thesis presents a novel learning-based framework, multi-level self-distillation (MSD), adapting a single-stream RGB tracker to the RGB-T setting through advanced training strategies rather than architectural changes. MSD integrates RGB and TIR data via a shared backbone guided by self-supervised contrastive and modality-gap alignment losses alongside supervised focal and modality-specific losses.
Extensive evaluations are performed to demonstrate that SBiDF and MSD provide performance superior to that of state-of-the-art tracking methods in terms of robust accuracy, simplified implementation, and enhanced computational efficiency, making them highly practical for real-world applications
Facial Attractiveness Prediction Using a Single and Multi-Task Vision Transformer Framework
No full textFacial attractiveness prediction is a challenging and inherently subjective task in computer vision, with applications spanning social media, cosmetic technology, and aesthetic medicine. While convolutional neural networks (CNNs) have driven significant advances in this area, recent developments in transformer-based architectures, such as the Vision Transformer (ViT), offer new opportunities by capturing global feature relationships and long-range dependencies within images. This thesis explores the use of Vision Transformers for predicting facial attractiveness on the SCUT-FBP5500 dataset, where beauty scores are computed from the average ratings of multiple human annotators. The task is formulated as a regression problem to predict continuous attractiveness scores. To enhance the learned feature representations, a multi-task learning framework is introduced, jointly performing gender and ethnicity classification alongside beauty prediction. The methodology includes systematic image preprocessing, transfer learning with a ViT pretrained on large-scale facial recognition data, and fine-tuning for both primary and auxiliary tasks. Model performance is evaluated using PC, MAE, and RMSE for regression and classification accuracy for auxiliary tasks. Comparative experiments with CNN-based baselines demonstrate that transformer architectures capture more holistic and subtle aesthetic cues, resulting in improved prediction consistency. Experimental results show that the proposed ViT-based approach achieves superior accuracy and robustness compared to conventional CNNs, even with limited training data. These findings highlight the potential of our Vision Transformers as an effective and data-efficient alternative for facial aesthetic analysis. The thesis concludes by emphasizing the value of multi-task learning in enriching feature representations and encourages future research toward interpretable and scalable beauty prediction systems
Developing Computer Vision-Based Digital Twin for Vegetation Management near Power Distribution Networks
No full textThe maintenance of power distribution lines is critically challenged by vegetation encroachment, posing significant risks to the reliability and safety of power utilities. Traditional manual inspection methods are resource-intensive and lack the precision required for effective and proactive maintenance. This paper presents an automated, accurate, and efficient approach to vegetation management near power lines by leveraging advancements in LiDAR as a remote sensing technology and deep learning algorithms. The RandLA-Net model is employed for semantic segmentation of large-scale point clouds to accurately identify vegetation, poles, and power lines. A comprehensive sensitivity analysis is conducted to optimize the model’s hyperparameters, enhancing segmentation accuracy. Post-processing techniques, including clustering and rule-based thresholding, are applied to refine the semantic segmentation results. Proximity detection is applied using spatial queries based on a KDTree structure to assess potential risks of vegetation near power lines. Furthermore, a digital twin of the power distribution network and surrounding trees is developed by integrating 3D object registration and surface generation, enriching it with semantic attributes and incorporating it into City Information Modeling (CIM) systems. This framework demonstrates the potential of remote sensing data integration for efficient environmental monitoring in urban infrastructure. The results of the case study on the Toronto-3D dataset demonstrate the computational efficiency and accuracy of the proposed method, presenting a promising solution for power utilities in proactive vegetation management and infrastructure planning. The optimized full 9-class model achieved an overall accuracy of 96.90% and IoU scores of 97.05% for vegetation, 88.09% for power lines, and 82.33% for poles, supporting comprehensive digital twin creation. An auxiliary 4-class model further improved targeted performance, with IoUs of 99.55% for vegetation, 88.79% for poles, and 87.18% for power lines
Domains of Wheelchair Users’ Socio-Emotional Experiences: Design Insights from a Scoping Review
Get PDFBackground
Physical accessibility is not the only concern for wheelchair users (WUs); they also face barriers to social presence, such as challenges in social engagement and negative stereotypes. Identifying key domains in the literature that impact their social and emotional experiences is essential to addressing these issues.
Objective
This scoping review sought to explore the key domains of WUs' socio-emotional experiences, as a foundation for providing design-oriented insights to enhance their social presence.
Methods
A literature search was conducted using the Web of Science, PubMed, Scopus, and PsycINFO databases, along with a manual search of three relevant journals. Articles in English, based on original empirical studies that focused on the socio-emotional experiences of adult WUs (>18), were included.
Results
Of the 48 articles included, most were from Canada (n = 11), Sweden (n = 9), the U.S. (n = 7), and the U.K. (n = 6), with limited studies from other countries. Among the six domains explored, Independence & Autonomy (26 %) was the most frequently reported, while Self-Identity & Body Image (9 %) and Social Stigma & Discrimination (5 %) were the least. Three interconnected themes emerged to guide design insights: Theme I – Foundations: Autonomy & Control, Theme II – Connections: Social Participation & Support, and Theme III – Reflection: Self- & Social-Identity.
Conclusion
While independence and agency are key concerns, little research has focused on perceptual issues like self- and social-identity, often highlighted in the media. This area can be refined by recognizing the crucial role of design in aesthetically shaping WUs' social representation in public settings
The dynamics of affective experiences with wheelchair use during rehabilitation: A qualitative study through physiotherapists' perspectives
No full textThe interaction between users and mobility aids, including emotional attachment and functional expectations, influences their perceptions and decisions on acceptance and continued use during rehabilitation. Tracking interactions during rehabilitation helps identify key intervention points, leading to effective therapeutic relationships and user-centered mobility aid designs. This study aims to track the dynamics of affective experiences (DAE) of wheelchair users (WUs) during a planned rehabilitation timeframe and recommend how to manage these dynamics. To this end, initially, the product experience framework was applied for the development of interview guidelines and analysis. Next, adopting a qualitative approach, semi-structured, in-depth interviews with 12 experienced physiotherapists were conducted in Iran. Transcripts were then analyzed using a thematic analysis framework to identify themes. A total of three themes have been identified which include: 1) Coping in Using the Wheelchair, 2) Reluctant Acceptance of the Wheelchair: Adjusting to the New Normal, and 3) Approaching Recovery: Challenges in Over-reliance. Additionally, two diagrams illustrating the dynamics of the affective experience of WUs and its influencing factors during rehabilitation have been provided. This study shows that the affective experience of WUs is not static and changes through various stages of rehabilitation. This dynamic is influenced by factors of emotional and functional importance, both of which often grow after initial resistance but follow varied patterns. However, emotional attachment can sometimes lead to over-reliance even after recovery, posing challenges in the rehabilitation. Physiotherapists can help balance this attachment, influencing users' affective experiences with their wheelchairs