Research Articles (Computer Science)

Permanent URI for this collectionhttp://hdl.handle.net/2263/1695

Browse

Recent Submissions

Now showing 1 - 20 of 245
  • Item
    Misinformation detection : a review for high and low resource languages
    (Universitas Bina Darma, 2024-12) Rananga, Seani; Modupe, Abiodun; Isong, Abiodun; Marivate, Vukosi; vukosi.marivate@up.ac.za
    The rapid spread of misinformation on platforms like Twitter, and Facebook, and in news headlines highlights the urgent need for effective ways to detect it. Currently, researchers are increasingly using machine learning (ML) and deep learning (DL) techniques to tackle misinformation detection (MID) because of their proven success. However, this task is still challenging due to the complexity of deceptive language, digital editing tools, and the lack of reliable linguistic resources for non-English languages. This paper provides a comprehensive analysis of relevant research, providing insights into advanced techniques for MID. It covers dataset assessments, the importance of using multiple forms of data (multimodality), and different language representations. By applying the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) methodology, the study identified and analyzed literature from 2019 to 2024 across five databases: Google Scholar, Springer, Elsevier, ACM, and IEEE Xplore. The study selected thirty-one papers and examined the effectiveness of various ML and DL approaches with a focal point on performance metrics, datasets, and false or misleading information detection challenges. The findings indicate that most current MID models are heavily dependent on DL techniques, with approximately 81% of studies preferring these over traditional ML methods. In addition, most studies are text-based, with much less attention given to audio, speech, images, and videos. The most effective models are mainly designed for highresource languages, with English datasets being the most used (67%), followed by Arabic (14%), Chinese (11%), and others. Less than 10% of the studies focus on low-resource languages (LRLs). Therefore, the study highlighted the need for robust datasets and interpretable, scalable MID models for LRLs. It emphasizes the critical need to prioritize and advance MID research for LRLs across all data types, including text, audio, speech, images, videos, and multimodal approaches. This study aims to support ongoing efforts to combat misinformation and promote a more informed understanding of underresourced African languages.
  • Item
    Optimizing power allocation for URLLCD2D in 5G networks with Rician fading channel
    (PeerJ Inc., 2025-02) Muhammad, Owais; Jiang, Hong; Bilal, Muhammad; Mushtaq, Muhammad Umer
    The rapid evolution of wireless technologies within the 5G network brings significant challenges in managing the increased connectivity and traffic of mobile devices. This enhanced connectivity brings challenges for base stations, which must handle increased traffic and efficiently serve a growing number of mobile devices. One of the key solutions to address these challenges is integrating device-to-device (D2D) communication with ultra-reliable and low-latency communication (URLLC). This study examines the impact of the Rician fading channel on the performance of D2D communication under URLLC. It addresses the critical problem of optimizing power allocation to maximize the minimum data rate in D2D communication. A significant challenge arises due to interference issues, as the problem of maximizing the minimum data rate is non-convex, which leads to high computational complexity. This complexity makes it difficult to derive optimal solutions efficiently. To address this challenge, we introduce an algorithm that is based on derivatives to find the optimal power allocation. Comparisons are made with the branch and bound (B&B) algorithm, heuristic algorithm, and particle swarm optimization (PSO) algorithm. Our proposed algorithm improves power allocation performance and also achieves faster execution with lower computational complexity compared to the B&B, PSO, and heuristic algorithms.
  • Item
    Advances in energy harvesting for sustainable wireless sensor networks : challenges and opportunities
    (MDPI, 2025-03) Mushtaq, Muhammad Umer; Venter, H.S. (Hein); Singh, Avinash; Owais, Muhammad; mu.mushtaq@up.ac.za
    Energy harvesting wireless sensor networks (EH-WSNs) appear as the fundamental backbone of research that attempts to expand the lifespan and efficiency of sensor networks positioned in resource-constrained environments. This review paper provides an in-depth examination of latest developments in this area, highlighting the important components comprising routing protocols, energy management plans, cognitive radio applications, physical layer security (PLS), and EH approaches. Across a well-ordered investigation of these features, this article clarifies the notable developments in technology, highlights recent barriers, and inquires avenues for future revolution. This article starts by furnishing a detailed analysis of different energy harvesting methodologies, incorporating solar, thermal, kinetic, and radio frequency (RF) energy, and their respective efficacy in non-identical operational circumstances. It also inspects state-of-the-art energy management techniques aimed at optimizing energy consumption and storage to guarantee network operability. Moreover, the integration of cognitive radio into EH-WSNs is acutely assessed, highlighting its capacity to improve spectrum efficiency and tackle associated technological problems. The present work investigates ground-breaking methodologies in PLS that uses energy-harvesting measures to improve the data security. In this review article, these techniques are explored with respect to classical encryption and discussed from network security points of view as well. The assessment furthers criticizes traditional routing protocols and their significance in EH-WSNs as well as the balance that has long been sought between energy efficiency and security in this space. This paper closes with the importance of continuous research to tackle existing challenges and to leverage newly avail- able means as highlighted in this document. In order to adequately serve the increasingly changing requirements of EH-WSNs, future research will and should be geared towards incorporating AI techniques with some advanced energy storage solutions. This paper discusses the integration of novel methodologies and interdisciplinary advancements for better performance, security, and sustainability for WSNs.
  • Item
    Cognitive strategies for UAV trajectory optimization : ensuring safety and energy efficiency in real-world scenarios
    (Elsevier, 2025-03) Mushtaq, Muhammad Umer; Venter, H.S. (Hein); Muhammad, Owais; Shafique, Tamoor; Awwad, Fuad A.; Ismail, Emad A.A.; mu.mushtaq@up.ac.za
    Many sectors in aerial transportation use unmanned aircraft vehicles (UAVs) extensively. This becomes even more challenging in complex environments where not only it is required to avoid obstacles, but it also must be maintained for a prolonged period of time. This paper presents a novel approach to increase UAV autonomy through safe and efficient flight trajectory design. An optimization problem is formulated with external and internal safety constraints, and traversing collision free paths. The proposed work offers an energy efficient RRT algorithm, which is used to assess multiple trajectory alternatives. The simulation results confirm the achieved performance in finding the optimal energy path while obeying to the safety constraint. The data and performance metrics, show the system operated in a safe and energy efficient manner. This work provides a unified framework for UAV trajectory planning that guarantees a trade-off between safety and energy efficiency.
  • Item
    Examination of customized questioned digital documents
    (Wiley, 2025-03) Adedayo, Oluwasola Mary; Olivier, Martin S.
    With the increasing trend of digitization of business processes and personal communication across the globe, digital documents of intrinsic value continue to be created. Whereas the questioned document examination (QDE) field of forensic science deals with the examination of “physical” documents potentially disputed in a court of law, there are no developed approaches for handling questioned digital documents (QDDs). Although techniques that address related problems such as identifying document types and image forensics exist, concrete strategies for analyzing questioned “digital” documents still need to be developed. This paper focuses on developing methods to examine QDDs that are customized from a database, due to the versatile use of customized documents in many areas. As a basis for our approach, we make the case for the need to develop analysis techniques for a digital counterpart of QDE which we term Questioned Digital Document Examination (QDDE). We posit that there is a benefit in considering digital aspects of forensic science disciplines where the questions answered by the discipline are clear, from a digital perspective. The paper describes some of the aspects that can be considered in the domain of question digital document examination. In designing methods for QDDE, we discuss the process of document recreation and describe the feasibility of our recreation process in different scenarios. Our experiments show that an alternative approach of considering digital aspects from a well-defined physical domain is worthwhile. It also supports the practical application of our approach in examining documents customized from a database.
  • Item
    Multi-objective evolutionary neural architecture search for recurrent neural networks
    (Springer, 2024-06-18) Booysen, Reinhard; Bosman, Anna; anna.bosman@up.ac.za
    Artificial neural network (NN) architecture design is a nontrivial and time-consuming task that often requires a high level of human expertise. Neural architecture search (NAS) serves to automate the design of NN architectures and has proven to be successful in automatically finding NN architectures that outperform those manually designed by human experts. NN architecture performance can be quantified based onmultiple objectives,which include model accuracy and some NN architecture complexity objectives, among others. The majority of modern NAS methods that consider multiple objectives for NN architecture performance evaluation are concerned with automated feed forward NN architecture design, which leaves multi-objective automated recurrent neural network (RNN) architecture design unexplored. RNNs are important for modeling sequential datasets, and prominent within the natural language processing domain. It is often the case in real world implementations of machine learning and NNs that a reasonable trade-off is accepted for marginally reduced model accuracy in favour of lower computational resources demanded by the model. This paper proposes a multi-objective evolutionary algorithm-based RNN architecture search method. The proposed method relies on approximate network morphisms for RNN architecture complexity optimisation during evolution. The results show that the proposed method is capable of finding novel RNN architectures with comparable performance to state-of-the-art manually designed RNN architectures, but with reduced computational demand.
  • Item
    Data fingerprinting and visualization for AI-enhanced cyber-defence systems
    (Institute of Electrical and Electronics Engineers, 2024-10) Klopper, Christiaan; Eloff, Jan H.P.; u26305667@tuks.co.za
    Artificial intelligence (AI)-assisted cyber-attacks have evolved to become increasingly successful in every aspect of the cyber-defence life cycle. For example, in the reconnaissance phase, AI-enhanced tools such as MalGAN can be deployed. The attacks launched by these types of tools automatically exploit vulnerabilities in cyber-defence systems. However, existing countermeasures cannot detect the attacks launched by most AI-enhanced tools. The solution presented in this paper is the first step towards using data fingerprinting and visualization to protect against AI-enhanced attacks. The AIECDS methodology for the development of AI-Enhanced Cyber-defense Systems was presented and discussed. This methodology includes tasks for data fingerprinting and visualization. The use of fingerprinted data and data visualization in cyber-defense systems has the potential to significantly reduce the complexity of the decision boundary and simplify the machine learning models required to improve detection efficiency, even for malicious threats with minuscule sample datasets. This work was validated by showing how the resulting fingerprints enable the visual discrimination of benign and malicious events as part of a use case for the discovery of cyber threats using fingerprint network sessions.
  • Item
    Fine-tuning retrieval-augmented generation with an auto-regressive language model for sentiment analysis in financial reviews
    (MDPI, 2024-12) Mathebula, Miehleketo; Modupe, Abiodun; Marivate, Vukosi; miehleketo.mathebula@tuks.co.za
    Sentiment analysis is a well-known task that has been used to analyse customer feedback reviews and media headlines to detect the sentimental personality or polarisation of a given text. With the growth of social media and other online platforms, like Twitter (now branded as X), Facebook, blogs, and others, it has been used in the investment community to monitor customer feedback, reviews, and news headlines about financial institutions’ products and services to ensure business success and prioritise aspects of customer relationship management. Supervised learning algorithms have been popularly employed for this task, but the performance of these models has been compromised due to the brevity of the content and the presence of idiomatic expressions, sound imitations, and abbreviations. Additionally, the pre-training of a larger language model (PTLM) struggles to capture bidirectional contextual knowledge learnt through word dependency because the sentence-level representation fails to take broad features into account. We develop a novel structure called language feature extraction and adaptation for reviews (LFEAR), an advanced natural language model that amalgamates retrieval-augmented generation (RAG) with a conversation format for an auto-regressive fine-tuning model (ARFT). This helps to overcome the limitations of lexicon-based tools and the reliance on pre-defined sentiment lexicons, which may not fully capture the range of sentiments in natural language and address questions on various topics and tasks. LFEAR is fine-tuned on Hellopeter reviews that incorporate industry-specific contextual information retrieval to show resilience and flexibility for various tasks, including analysing sentiments in reviews of restaurants, movies, politics, and financial products. The proposed model achieved an average precision score of 98.45%, answer correctness of 93.85%, and context precision of 97.69% based on Retrieval-Augmented Generation Assessment (RAGAS) metrics. The LFEAR model is effective in conducting sentiment analysis across various domains due to its adaptability and scalable inference mechanism. It considers unique language characteristics and patterns in specific domains to ensure accurate sentiment annotation. This is particularly beneficial for individuals in the financial sector, such as investors and institutions, including those listed on the Johannesburg Stock Exchange (JSE), which is the primary stock exchange in South Africa and plays a significant role in the country’s financial market. Future initiatives will focus on incorporating a wider range of data sources and improving the system’s ability to express nuanced sentiments effectively, enhancing its usefulness in diverse real-world scenarios.
  • Item
    Training feedforward neural networks with Bayesian hyper-heuristics
    (Elsevier, 2025-01) Schreuder, Arné; Bosman, Anna Sergeevna; Engelbrecht, Andries P.; Cleghorn, Christopher W.; an.schreuder@up.ac.za
    The process of training feedforward neural networks (FFNNs) can benefit from an automated process where the best heuristic to train the network is sought out automatically by means of a high-level probabilistic-based heuristic. This research introduces a novel population-based Bayesian hyper-heuristic (BHH) that is used to train feedforward neural networks (FFNNs). The performance of the BHH is compared to that of ten popular low-level heuristics, each with different search behaviours. The chosen heuristic pool consists of classic gradient-based heuristics as well as meta-heuristics (MHs). The empirical process is executed on fourteen datasets consisting of classification and regression problems with varying characteristics. The BHH is shown to be able to train FFNNs well and provide an automated method for finding the best heuristic to train the FFNNs at various stages of the training process.
  • Item
    ChatGPT as a text annotation tool to evaluate sentiment analysis on South African financial institutions
    (Institute of Electrical and Electronics Engineers, 2024-09) Mathebula, Miehleketo; Modupe, Abiodun; Marivate, Vukosi
    Social media platforms play a significant role in analyzing customer perceptions of financial products and services in today’s culture. These platforms facilitate the immediate and in-depth sharing of thoughts and experiences, offering valuable insights into consumer behaviour. Any customer looking for such a service would surf the internet for reviews and ratings before making a decision, which usually influences their ultimate pick. Feedback and suggestions from friends, family, and coworkers improve customer experiences. Customer reviews play a crucial role in shaping the reputation and profitability of businesses and products offered by financial institutions, often serving as the final assessment of quality and satisfaction during decision-making. Therefore, it is paramount for decision-makers to carefully evaluate customer feedback and understand the sentiment expressed in a given piece of text, which could lead to equity trading, and credit market assessment, and offer invaluable insights that boost the financial performance of the institution. Previous research has used human-annotated text, such as lexicon-based methods, to train machine learning models for sentiment analysis, but the approach did not capture the full range of structure and semantic relationships in natural language. Therefore, our research aims to develop a more comprehensive and accurate sentiment analysis model using advanced natural language processing techniques that could answer questions on various subjects and tasks. To do this, we first crawled customer reviews on Hellopeter, a popular review site, and financial data on the top five financial institutions listed on the Johannesburg Stock Exchange (JSE) in South Africa. After that, we used OpenAI’s ChatGPT as a zero-short learning model to generate human-like annotation tools for different sentiment tasks. The OpenAI ChatGPT feature vector was subsequently fed into BERT, BiLSTM, and a SoftMax function to detect and identify the sentiment of a given sentence. Lastly, we use feature vectors with oversampling methods to address the imbalanced data dilemma and visualise the contribution features of the given piece of text for the customer reviewers. The experiments demonstrated that the method performed as well as or better than the latest and most effective methods on the tested datasets, yielding comparable results. When OpenAI’s ChatGPT was combined with pre-trained BERT and BiLSTM models, it did better overall, with an average score of 98.9%, an F1-measure of 97.7%, and an AUC of 91.90% when oversampling was used. The traditional lexicon-based model got an 86.68% score using SVM and logistic regression and an AUC of 91.90%. The study shows the exceptional performance of OpenAI ChatGPT in detecting the emotional tone or polarity of a given sentence in a customer review, which helps with annotation and understanding the sentiment analysis of an event and how it influences decisions and outcomes. In conclusion, these results underscore the significant advantages of incorporating customer sentiment analysis into financial analysis and decision-making processes as a valuable tool for understanding and prioritizing customer needs and preferences.
  • Item
    On the transparency of large AI models
    (Cell Press, 2023-07) Wang, Wanying; Wang, Ge; Marivate, Vukosi; Hufton, Andrew L.
    As large AI models demonstrate increasingly human-like performance on complex tasks, many scientists are developing or adapting these models to empower their research and applications. Because of the substantial costs involved in building, training, and running large AI models, closedsource models can often offer performance that cannot be matched by open-source counterparts, making them tempting tools for researchers even if they are not transparent or accessible according to conventional academic standards. Moreover, even researchers who are developing their own AI models may face special challenges when trying to publish their work in an open and reproducible manner. In particular, the very large datasets required to train AI models often come with special challenges that make them inherently hard to share—ranging from sheer size to tricky copyright and privacy issues. In this editorial, we share some insights and tips that we hope will help researchers in this field understand our journal’s policies and prepare submissions for the journal.
  • Item
    Denoising diffusion post-processing for low-light image enhancement
    (Elsevier, 2024-12) Panagiotou, Savvas; Bosman, Anna; u17215286@tuks.co.za
    Low-light image enhancement (LLIE) techniques attempt to increase the visibility of images captured in low-light scenarios. However, as a result of enhancement, a variety of image degradations such as noise and color bias are revealed. Furthermore, each particular LLIE approach may introduce a different form of flaw within its enhanced results. To combat these image degradations, post-processing denoisers have widely been used, which often yield oversmoothed results lacking detail. We propose using a diffusion model as a post-processing approach, and we introduce Low-light Post-processing Diffusion Model (LPDM) in order to model the conditional distribution between under-exposed and normally-exposed images. We apply LPDM in a manner which avoids the computationally expensive generative reverse process of typical diffusion models, and post-process images in one pass through LPDM. Extensive experiments demonstrate that our approach outperforms competing post-processing denoisers by increasing the perceptual quality of enhanced low-light images on a variety of challenging low-light datasets. Source code is available at https://github.com/savvaki/LPDM.
  • Item
    Using distributed ledger technology for digital forensic investigation purposes on tendering projects
    (Springer, 2023-03-09) Ramazhamba, T.P.; Venter, H.S. (Hein); hein.venter@up.ac.za
    The South African Local Government (SALG) uses the tendering system to procure goods and services. Some of these tendering projects are aimed at promoting socio-economic and industrial policies. Hence, the tendering system used by SALG should be fair, transparent, competitive, cost-effective, equitable, and free from corruption. However, the mismanagement of the tendering system might lead to interruption of operations, late service delivery, rising costs, and most importantly, fraud and corruption. The use of paperwork to share project information might lead to the mismanagement of the tendering project because it might contribute towards illicit altering of project information during the process. The purpose of this study is to develop a Blockchain prototype that might be used to securely share project information with all the parties interested in the tendering project. It is recommended that the adoption of the proposed solution will enable various organisations to have access to real-time data, allowing them to have access to the entire project history regardless of their geographical location. Access to real-time data would promote real-time auditing and digital forensic investigations because both auditors and investigators will have access to credible digital evidence or project information of their interest in real-time.
  • Item
    Exploring COVID-19 public perceptions in South Africa through sentiment analysis and topic modelling of Twitter posts
    (Learning Information Networking and Knowledge (LINK) Centre, Graduate School of Public and Development, 2023) Kekere, Temitope; Marivate, Vukosi; Hattingh, Maria J. (Marie)
    The narratives shared on social media during a health crisis such as COVID-19 reflect public perceptions of the crisis. This article provides findings from a study of the perceptions of South African citizens regarding the government’s response to the COVID-19 pandemic from March to May 2020. The study analysed Twitter data from posts by government officials and the public in South Africa to measure the public’s confidence in how the government was handling the pandemic. A third of the tweets dataset was labelled using valence aware dictionary and sentiment reasoner (VADER) lexicons, forming the training set for four classical machinelearning algorithms—logistic regression (LR), support vector machines (SVM), random forest (RF), and extreme gradient boosting (XGBoost)—that were employed for sentiment analysis. The effectiveness of these classifiers varied, with error rates of 17% for XGBoost, 14% for RF, and 7% for both SVM and LR. The best-performing algorithm (SVM) was subsequently used to label the remaining two-thirds of the tweet dataset. In addition, the study used, and evaluated the effectiveness of, two topic-modelling algorithms—latent dirichlet allocation (LDA) and non-negative matrix factorisation (NMF)—for classification of the most frequently occurring narratives in the Twitter data. The better-performing of these two algorithms, NMF, identified a prevalence of positive narratives in South African public sentiment towards the government’s response to COVID-19.
  • Item
    Neural network crossover in genetic algorithms using genetic programming
    (Springer, 2024-06) Pretorius, Kyle; Pillay, Nelishia; u16234805@tuks.co.za
    The use of genetic algorithms (GAs) to evolve neural network (NN) weights has risen in popularity in recent years, particularly when used together with gradient descent as a mutation operator. However, crossover operators are often omitted from such GAs as they are seen as being highly destructive and detrimental to the performance of the GA. Designing crossover operators that can effectively be applied to NNs has been an active area of research with success limited to specific problem domains. The focus of this study is to use genetic programming (GP) to automatically evolve crossover operators that can be applied to NN weights and used in GAs. A novel GP is proposed and used to evolve both reusable and disposable crossover operators to compare their efficiency. Experiments are conducted to compare the performance of GAs using no crossover operator or a commonly used human designed crossover operator to GAs using GP evolved crossover operators. Results from experiments conducted show that using GP to evolve disposable crossover operators leads to highly effectively crossover operators that significantly improve the results obtained from the GA.
  • Item
    A genetic programming approach to the automated design of CNN models for image classification and video shorts creation
    (Springer, 2024-03) Kapoor, Rahul; Pillay, Nelishia; u16034130@tuks.co.za
    Neural architecture search (NAS) is a rapidly growing field which focuses on the automated design of neural network architectures. Genetic algorithms (GAs) have been predominantly used for evolving neural network architectures. Genetic programming (GP), a variation of GAs that work in the program space rather than a solution space, has not been as well researched for NAS. This paper aims to contribute to the research into GP for NAS. Previous research in this field can be divided into two categories. In the first each program represents neural networks directly or components and parameters of neural networks. In the second category each program is a set of instructions, which when executed, produces a neural network. This study focuses on this second category which has not been well researched. Previous work has used grammatical evolution for generating these programs. This study examines canonical GP for neural network design (GPNND) for this purpose. It also evaluates a variation of GP, iterative structure-based GP (ISBGP) for evolving these programs. The study compares the performance of GAs, GPNND and ISBGP for image classification and video shorts creation. Both GPNND and ISBGP were found to outperform GAs, with ISBGP producing better results than GPNND for both applications. Both GPNND and ISBGP produced better results than previous studies employing grammatical evolution on the CIFAR-10 dataset.
  • Item
    Combating hate : how multilingual transformers can help detect topical hate speech
    (Easychair, 2023) Srikissoon, Trishanta; Marivate, Vukosi; vukosi.marivate@cs.up.ac.za
    Automated hate speech detection is important to protecting people’s dignity, online experiences, and physical safety in Society 5.0. Transformers are sophisticated pre-trained language models that can be fine-tuned for multilingual hate speech detection. Many studies consider this application as a binary classification problem. Additionally, research on topical hate speech detection use target-specific datasets containing assertions about a particular group. In this paper we investigate multi-class hate speech detection using target-generic datasets. We assess the performance of mBERT and XLM-RoBERTA on high and low resource languages, with limited sample sizes and class imbalance. We find that our fine-tuned mBERT models are performant in detecting gender-targeted hate speech. Our Urdu classifier produces a 31% lift on the baseline model. We also present a pipeline for processing multilingual datasets for multi-class hate speech detection. Our approach could be used in future works on topically focused hate speech detection for other low resource languages, particularly African languages which remain under-explored in this domain.
  • Item
    Regularised feed forward neural networks for streamed data classification problems
    (Elsevier, 2024-07) Ellis, Mathys; Bosman, Anna Sergeevna; Engelbrecht, Andries P.
    Streamed data classification problems (SDCPs) require classifiers to not just find the optimal decision boundaries that describe the relationships within a data stream, but also to adapt to changes in the decision boundaries in real-time. The requirement is due to concept drift, i.e., incorrect classifications caused by decision boundaries changing over time. Changes include disappearing, appearing or shifting decision boundaries. This article proposes an online learning approach for feed forward neural networks (FFNNs) that meets the requirements of SDCPs. The approach uses regularisation to dynamically optimise the architecture, and quantum particle swarm optimisation (QPSO) to dynamically adjust the weights. The learning approach is applied to a FFNN, which uses rectified linear activation functions, to form a novel SDCP classifier. The classifier is empirically investigated on several SDCPs. Both weight decay (WD) and weight elimination (WE) are investigated as regularisers. Empirical results show that using QPSO with no regularisation causes the classifier to completely saturate. However, using QPSO with regularisation makes the classifier efficient at dynamically adapting both its architecture and weights as decision boundaries change. Furthermore, the results favour WE over WD as a regulariser for QPSO.
  • Item
    Application of deep reinforcement learning in asset liability management
    (Elsevier, 2023-11) Wekwete, Takura Asael; Kufakunesu, Rodwell; Van Zyl, A.J. (Gusti); rodwell.kufakunesu@up.ac.za
    Asset Liability Management (ALM) is an essential risk management technique in Quantitative Finance and Actuarial Science. It aims to maximise a risk-taker’s ability to fulfil future liabilities. ALM is especially critical in environments of elevated interest rate changes, as has been experienced globally between 2021 and 2023. Traditional ALM implementation is still heavily dependent on the judgement of professionals such as Quants, Actuaries or Investment Managers. This over-reliance on human input critically limits ALM performance due to restricted automation, human irrationality and restricted scope for multi-objective optimisation. This paper addressed these limitations by applying Deep Reinforcement Learning (DRL), which optimises through trial, and error and continuous feedback from the environment. We defined the Reinforcement Learning (RL) components for the ALM application: the RL decision-making Agent, Environment, Actions, States and Reward Functions. The results demonstrated that DRL ALM can achieve duration-matching outcomes within 1% of the theoretical ALM at a 95% confidence level. Furthermore, compared to a benchmark weekly rebalancing traditional ALM regime, DRL ALM achieved superior outcomes of net portfolios which are, on average, 3 times less sensitive to interest rate changes. DRL also allows for increased automation, flexibility, and multi-objective optimisation in ALM, reducing the negative impact of human limitations and improving risk management outcomes. The findings and principles presented in this study apply to various institutional risk-takers, including insurers, banks, pension funds, and asset managers. Overall, DRL ALM provides a promising Artificial Intelligence (AI) avenue for improving risk management outcomes compared to the traditional approaches.
  • Item
    Technology-enhanced learning, data sharing, and machine learning challenges in South African education
    (MDPI, 2023-04-24) Combrink, Herkulaas MvE; Marivate, Vukosi; Masikisiki, Baphumelele; u29191051@tuks.co.za
    The objective of this paper was to scope the challenges associated with data-sharing governance for machine learning applications in education research (MLER) within the South African context. Machine learning applications have the potential to assist student success and identify areas where students require additional support. However, the implementation of these applications depends on the availability of quality data. This paper highlights the challenges in data-sharing policies across institutions and organisations that make it difficult to standardise data-sharing practices for MLER. This poses a challenge for South African researchers in the MLER space who wish to advance and innovate. The paper proposes viewpoints that policymakers must consider to overcome these challenges of data-sharing practices, ultimately allowing South African researchers to leverage the benefits of machine learning applications in education effectively. By addressing these challenges, South African institutions and organisations can improve educational outcomes and work toward the goal of inclusive and equitable education.