3. Electronic Theses and Dissertations (ETDs)

Permanent URI for this communityhttps://hdl.handle.net/10539/37773

For queries relating to content and technical issues, please contact IR specialists via this email address: openscholarship.library@wits.ac.za, Tel: 011 717 4652 or 011 717 1954

Browse

Search Results

Now showing 1 - 10 of 636

A study of the second-order dynamical systems for variational inequalities
(University of the Witwatersrand, Johannesburg, 2025-05) Ranoto, Tumelo; Izuchukwu, Chinedu
The purpose of this paper is to study the variational inequality problem through a second order dynamical system. The dynamical system is a second-order dynamical system that involves the asymptotic vanishing of damping and Hessian-damping terms. Our approach primarily uses the asymptotic dynamics of the inertial system and the geometric characteristics of the damping factors. The asymptotic vanishing damping is directly linked to the method of the Nesterov accelerated gradient. The geometry of the Hessian damping term is in this form d dt(y(t)−w(t)), where y(t) is the generated trajectories by our proposed dynamical system, w(t) = PC(y(t)−Qy(t)) and Q is our operator which is strongly monotone. The Hessian term assists with the neutralisation of the oscillation. This combination forms a damping force that assists with controlling the convergence speed of dynamical systems, stabilizing the system, and mostly ensuring we have a controlled velocity over time. As a main result, we establish the existence and uniqueness of the generated trajectories using the Cauchy-Lipschitz theorem. We give two numerical non-Hilbert examples. We construct a Lyapunov function to achieve weak convergence, assuming that the underlying operator is Lipschitz continuous and monotone. By considering the strong monotonicity of the operator, we establish exponential convergence. Furthermore, we given two numerical examples to show the applicability of our results. Finally, we conduct numerical experiments to show the performance of our dynamical system. The results presented in this paper build upon and significantly enhance recent findings in the existing literature.
Impact of Educational Interventions on Mitigating Skills Shortages in the South African Mining Sector
(University of the Witwatersrand, Johannesburg, 2025) Nhlapo, Sinenhlanhla Randy; Oro, Oro Ufuo
fficient in mining operations. Furthermore, the educational curricula have not kept up with the technological advancements and rapid adoption of modern technologies, further contributing to the skills shortage challenge. This study recommends enhancing training programmes with experiential learning methods, aligning educational curricula with modern technologies, and increasing collaboration between mining companies and government to promote STEM education in schools. These strategies aim to build a future-ready workforce, ensuring the sector’s sustainability and competitiveness in an increasingly technology-driven environment.
Archiving Visibility Data Using Lossy Baseline-Dependent SVD Techniques
(University of the Witwatersrand, Johannesburg, 2025-06) Ramanyimi, Mukundi; Atemkeng, Marcellin; Makhathini, Sphesihle
Modern radio interferometer arrays, such as the MeerKAT [1], the Australian Square Kilometre Array Pathfinder (ASKAP) [2, 3], the Low Frequency Array (LOFAR) [4], the Murchison Widefield Array (MWA) [5, 6], and the upcoming Square Kilometre Array Observatory (SKAO) [7], generate large volumes of data due to their high temporal and spectral resolutions, large number of baseline configurations, and wide bandwidths. Managing these data volumes poses substantial challenges in terms of storage and processing. To address the growing costs, averaging techniques are widely used to reduce data sizes. However, averaging leads to signal loss in radio interferometric images, resulting in smeared or blurred source emissions and reduced source amplitudes. Moreover, the extent of this smearing is baseline-dependent, as the signal phase depends on baseline length. Specifically, longer baselines are more affected than shorter ones. This is addressed by Baseline Dependent Averaging (BDA), which applies variable averaging intervals - longer for shorter baselines and shorter for longer baselines. BDA achieves high data volume reduction since radio interferometers generally have more shorter baselines, which can be aggressively averaged with minimal smearing effects. However, BDA changes the time-frequency grid structure of the data, making it incompatible with the standard storage format in the field, the Measurement Set (MS). A promising approach to data compression was presented by Atemkeng et al. [8], who developed a compression technique based on Singular Value Decomposition (SVD). This approach exploits the inherent structure of raw visibility data, representing it as a low-rank matrix approximation where each component corresponds to a specific Fourier component of the sky distribution. By approximating the data with a reduced rank, the essential features of the original data can be captured using fewer components, effectively reducing data size. In this work, we build on the methods introduced by Atemkeng et al. [8], specifically evaluating the effectiveness of SVD in compressing large volumes of data while preserving image quality and data fidelity for long-term archival. Although our study focuses on the MeerKAT telescope, the approach can be adapted for use with any other radio telescope. Our findings demonstrate that for a bright point source (1 Jy), whether located at the phase centre or away from it, the data features can effectively be captured using a single component, recovering over 99.90% of the source amplitude and achieving a data size reduction of over 97%. For fields with multiple sources, the features can be fully captured using 3-4 components out of 24, recovering over 99.90% of the source amplitude for a source at the edge of the Field of View (FoV), which is around 1.1 deg for the MeerKAT at a frequency of 1.4 GHz. This results in a data size reduction of over 91%. Additionally, we found that the source or field direction does not impact SVD compression. On the other hand, the Signal to Noise Ratio (SNR) significantly affects SVD compression. For sources with low SNR or faint sources, all components are required to recover more than 97% of the source amplitude, making the compression ineffective. In such scenarios, it would be more advantageous to first denoise the data or to use BDA.
Cross-domain few-shot classification for remote sensing imagery
(University of the Witwatersrand, Johannesburg, 2025-04) Pillay, Christopher Wayne; Bau, Hairong
Deep learning has proven highly effective for scene classification tasks when substantial quantities of labelled data are accessible. However, performance decreases when applied to domains such as remote sensing which typically possess a limited quantity of labelled data across available datasets. Few-shot learning has been developed as one of the promising solutions to this problem. It has the ability to recognise new categories with minimal labelled examples, but it assumes that the training and testing data will exhibit identical feature distributions. This assumption is unrealistic in real-world contexts where data can originate from different domains and poses a challenge when a significant domain shift exists between the training and testing data. This dissertation aims to address these limitations by proposing the Cross-Domain Attention Network (CDAN). It is a network designed specifically to solve the issues that arise when there is a limited quantity of labelled data available and a significant domain shift exists between the training and testing data. The network proposed consists of a prototypical network as the base and three additions that contribute to the accurate scene classification of remote sensing imagery. Firstly, a cross-domain data augmentation technique is proposed with few-shot learning to reduce domain shift. The cross-domain data augmentation technique facilitates enhanced knowledge transfer between domains and increases the adaptation ability of the network, whereas few-shot learning reduces the network’s reliance on large labelled datasets. Secondly, a dynamic and focused attention module is proposed to improve discriminative capacity of the network by increasing the focus on important channels and spatial regions within images during training. Thirdly, an adaptive task aware loss is proposed to further enhance the network’s adaptive capacity by leveraging information in few-shot training tasks. Extensive experiments are carried out with different remote imaging classification datasets (RSICB, AID and NWPU-RESISC45) to prove that the proposed network alleviates concerns in a cross-domain few-shot (CDFS) classification setting.
What Are B-jets and What Can We Do With Them? A beauty-ful exploration of b-jets
(University of the Witwatersrand, Johannesburg, 2025-04) Nzuza, Wandile Siyamthanda; Kar, Deepak
In the Standard Model of particle physics, the bottom quark is one of the six quarks. It belongs to the third generation of quarks, alongside the top quark. The b-quark has a charge of - 1/3 of the elementary charge, making it negatively charged, and a mass of 4.2 GeV significantly greater than the up and down quarks, but smaller than that of the top quark. At the LHC, two protons are accelerated close to the speed of light in a head-on collision and as a result there are final states with thousands of hadrons. This means that there are jets everywhere. Jets are not fundamental objects observed in the final state but are a cone-like structure constructed to help deal with the large number of hadrons in final states. Jets containing b-hadrons are known as b-jets. The reason that final states with b-jets are of interest is because they result in much less ambiguous collider signatures which can assist in reducing the background significantly. The results presented are from various studies done using b-jets. We started off with a study of bottom-quark-philic semi-visible jets, followed by a measurement of b-jet cross-section as a function of missing transverse momentum using data collected by the ATLAS detector in Run 2 of the LHC and lastly two Performance Studies done of heavy-flavour tagging in ALICE and ATLAS experiment at the LHC.
Graduate employee adaption to organisational culture and performance
(University of the Witwatersrand, Johannesburg, 2025) Naude, Barend; Gobind, Jenika
This study outlined the integration of graduate employees into organisational culture and performance system settings, focusing on a South African retail organisation. This research explored the distinctive contributions that graduates, with contemporary academic knowledge and new insights, can offer to the dynamics, culture and operational performance of the organisation. This case study probed graduate integration, the influence of graduate integration on organisational culture in the medium term, performance indicators for the immediate to medium term, and the challenges and opportunities presented during the induction process. The central aim of the inquiry was an assessment of how organisational leadership, mentorship and onboarding processes facilitate or hinder assimilation into existing corporate frameworks. The study portrays the dual role of graduates as disruptors and enhancers of organisational culture, underlining the potential contributions of graduates in driving innovation, process improvement and collaboration, while gaps in soft skills and cultural misalignment act as potential barriers. These findings provide recommendations that can be acted upon by corporate leaders, HR professionals and policymakers in their quest to optimally integrate and develop graduate talent while reinforcing a culture of inclusion, continuous learning and performance excellence. The research contributed a deeper insight into workforce development strategies and their implications for organisational growth and adaptability within a dynamic and competitive business environment.
Engaging Collections: Libetshu in the Van Warmelo Collection, Ditsong National Museum of Cultural History, South Africa
(University of the Witwatersrand, Johannesburg, 2025) Nzimela, Yola Siyamthanda; Namono, Catherine
This thesis broadly examines curatorial practice of African ritual material culture within museum and related institutional collections in South Africa. Specifically, I focus on divination material known as libetshu, or Mankgwanyana, heir looms, in the Nicolaas van Warmelo collection, the Ditsong National Museum of Cultural History, Pretoria, South Africa. Although many communities from which ritual objects such as these divination materials originate still revere them as sacred, there is a lack of meaningful inclusion of source communities in curatorial decision-making mainly due to elitist or Eurocentric perspectives on culture due to colonial legacies and/or where curatorial practices reside with Western trained professionals. Globally and locally, limited accurate archival contextual data about such divination material constrains connecting the objects to source communities. This thesis therefore sought to enhance understanding of libetshu /Mankgwanyana as ritually activated divination material, to encourage appropriate curation that takes cognisance of their spiritual significance for communities that hold them sacred. Through use-wear analysis, semi-structured interviews with traditional practitioners and curators, and a curatorial justice framework, I show that libetshu / Mankgwanyana are divination materials used in ritual performances, are spiritual and require specific handling and curation. I find that whilst co-curation with source communities and stakeholders is essential, a blanket approach for ritual objects is insufficient as each divination material belongs to different ancestral lineages. These findings highlight the importance of epistemic restitution and demonstrate the effectiveness of a curatorial justice approach that promotes sensitive, respectful, and appropriate curatorial practices cognisant of the integrity of divination materials such as libetshu / Mankgwanyana in collections.
Leadership Challenges in a South African Sector Education and Training Authority (SETA)
(University of the Witwatersrand, Johannesburg, 2025) Mudau, Fhatuwani; Matshabaphala, Manamela
Since the dawn of democracy in 1994, South Africa has realised significant changes across different spectrums, with leadership being one of those. Nelson Mandela became the first leader to be democratically elected as president of the Republic South Africa and led the entire nation irrespective of gender, race, ethnic group, and socio-economic status. Leadership is a highly complex topics as it has many dynamic challenges and opportunities. This research study assesses the leadership challenges in a South African Sector Education and Training Authority. There are twenty-one (21) Sector Education and Training Authorities (SETAs). Each of these SETAs is designated to facilitate skills development in their respective economic clusters as outlined by the Skills Development Act, No. 97 of 1998 as amended (the Act). The research study explored different leadership theories to assess the leadership challenges in the sector education and training authority, including situational leadership, laissez-faire leadership, servant leadership, democratic leadership, and exemplary leadership. Ultimately, the exemplary leadership resonated with the topic at hand as well as the problem statement and research gap identified. To explore this further, the researcher drew on the framework of Kouzes and Posner (2012) about the Five Practices of Exemplary Leadership. The framework comprises five elements of exemplary leadership, which are model the way, inspire a shared vision, challenge the process, enable others to act, and encourage the heart. These elements of exemplary leadership were examined by developing a measurement instrument to explore the leadership challenges in the sector education and training authority. The measurement instrument together with quantitative methods were utilised as tools to gather data and gain insight into the challenges at hand from 200 sampled employees of the sector ii education and training authority. A non-probability convenience sampling method was utilised. The results revealed that there was a significant relationship between inspiring a shared vision and leadership challenges. Results revealed that there was a significant relationship between encouraging the heart and leadership challenges. The other three hypotheses were insignificant.
Parental perceptions of gamification and educational outcomes in public schools in Gauteng
(University of the Witwatersrand, Johannesburg, 2024) Miswe, Fulufhelo
A Bayesian Approach to Maximise Photovoltaic System Output
(University of the Witwatersrand, Johannesburg, 2025-05) Noel, Keanu; Ajoodha, Ritesh
This thesis addresses the critical issue of optimising photovoltaic (PV) system output, an essential objective in the pursuit of efficient and scalable renewable energy solutions. As global energy demands rise and concerns over climate change intensify, solar power has emerged as a leading solution for sustainable electricity generation. However, the performance of PV systems is highly sensitive to environmental factors which can vary significantly across seasons and geographical locations. These fluctuations create a complex optimisation problem in determining the most effective system configuration that can dynamically adapt to seasonal and regional variations in solar potential. Traditional approaches often rely on fixed or rule-based models that do not adequately account for these variations, leading to suboptimal energy yields and the inefficient use of solar infrastructure. In this research, a Bayesian Network model is developed to learn the conditional dependencies between meteorological variables (such as solar irradiance, temperature, and wind speed) and PV system configuration parameters (tilt angle, orientation, inverter properties). By using a Bayesian approach, the developed model accommodates uncertainty and dynamically adjusts PV system tilt configurations to weather variations, aiming to maximise PV output. Score-based methods are employed to construct the network structure, and Maximum Likelihood Estimation (MLE) to determine the Conditional Probability Distributions (CPDs) of the network. Additionally, Maximum a Posteriori (MAP) estimation is applied to identify the optimal seasonal PV system tilt configurations in light of specific weather conditions. Key findings demonstrate the effectiveness of the model in optimising PV output by offering adaptive configuration strategies that respond to local seasonal meteorological patterns. This includes the superior performance of the Hill Climb Search algorithm compared to Simulated Annealing for structure learning, the utilisation of MAP Estimation for identifying optimal PV system tilt configurations under varying meteorological conditions, and the statistically significant advantage of dynamic configurations over fixed installations for enhancing PV system output. These results underscore the potential of Bayesian approaches for data-driven optimisation in renewable energy systems. This research provides a robust framework that enhances PV system performance and contributes to the growing body of knowledge on renewable energy optimisation through probabilistic modelling. Ultimately, this research presents a novel, data-driven methodology which informs the design and operation of more efficient PV systems, answering both the “so what” and “now what” in the context of sustainable energy advancements.

3. Electronic Theses and Dissertations (ETDs)

Browse

Filters

Settings

Sort By

Results per page

Search Results