Researcher: Linhle Mbombo, University of Venda
Supervisor: Professor Augustine Munagi and Professor Turgay Çelik, University of the Witwatersrand, Johannesburg

Machine learning (ML) algorithms is improving auto-mated tasks, using data to make predictions and solve clustering problems. The data that is used to fit the model is from institutions, organizations, etc. that have sensitive data. Such as personal, medical and financial data. The research seeks to bridge the security gab and design secure measures of preserving the privacy of data used in the process. Homomorphic encryption concept allows computations on encrypted data. The research uses Paillier algorithm for an encryption scheme.

Researcher: Makgoka Nkoana, University of the Witwatersrand, Johannesburg
Supervisor: Dr Farai Mlambo, University of the Witwatersrand, Johannesburg

Wavelet Neural Networks were inspired by neural networks as well as wavelet decomposition, and as such Zhang and Beneveniste sort to combine the two concepts. This particular network is often used in the financial industry for making forecasts related to the stock market and as such this Investigation sort to compare the results of stock market forecasts made several implementations of the Wavelet Neural Networks, with different activation functions. The main findings from this investigation found that the Shannon Wavelet Neural Network and the Gaussian Neural Network were the best performing followed by the Mexican Hat Neural Network.

Researcher: Seleme Shoky, University of Venda
Supervisor: Dr Farai Mlambo, University of the Witwatersrand, Johannesburg

Each year credit card fraud is growing significantly with the advancements of technology resulting in extreme losses to those affected. We build ML model to detect fraudulent activity in credit card transaction systems. The binary classifiers build are Neural Network and Random Forest. We used Random Forest for variable importance. The aim to develop an approach which will detect fraud with high recall score and low number of false positives(Sampling Techniques). We used ROC curves, confusion matrices and precision recall statistics to measure the performance of the models.

Researcher: Sindisiwe Zulu, University of the Witwatersrand, Johannesburg
Supervisor: Mr Rendani Mbuvha, University of the Witwatersrand, Johannesburg

Wind energy is seen as the next promising renewable energy to be used for future power generation. The stochastic wind behavior has resulted in the development of improved wind forecasting techniques. New techniques are required for wind speed forecasting. This research investigates Long Short Term Memory (LSTM) networks for 1 hour to 3 hours ahead forecasting of wind speed. In comparison with LSTM, the Multilayer Perceptron (MLP) model tests the LSTM model’s efficiency. From the results it is shown that the LSTM model outperformed the MLP model however, the differences in performance are not statistically significant. Best results were obtained from 1 hour ahead forecasting.

Researcher: Mamphaga Ratshilengo, University of Venda, Johannesburg
Supervisor: Dr Farai Mlambo, University of the Witwatersrand, Johannesburg

Wavelet is a wavy oscillation with width, scale, or magnitude that begin at zero point, rises, and decline to zero again [1]. Neural network (NN) is a series of algorithms aims in recognising the relations in a set of data through a process that characterises the work of human brain. NNs are applicable in various field, including speech recognition, hand-written digit recognition and when driving a car [2]. In WNNs, wavelets and NNs are combined into one thing [1].

Researcher: Elishua Ngoma, University of The Witwatersrand, Johannesburg
Supervisor: Professor Rod Alence, University of the Witwatersrand, Johannesburg

Over the past two decades Sino-Africa bilateral and multilateral relations have increased at a tremendous rate. Resulting in China becoming the largest trading partner of many African countries. China’s ever-growing economy and population has resulted in a strategic scramble for resources. Its unorthodox approach towards the procurement of these natural resources has resulted in it facing enormous global criticism, much of which is targeted at its perceived apathy towards human rights violations and corrupt governance. A dominant narrative within mainstream media is that china specifically targets ill governed states for trade and aid relations. This capstone project sought to find out if there was any credibility to these narratives

Researcher: Kgaugelo Mphahlele, University of The Witwatersrand, Johannesburg
Supervisor: Dr Farai Mlambo, University of the Witwatersrand, Johannesburg

Wavelet Neural Networks were inspired by neural networks as well as wavelet decomposition, and as such Zhang and Beneveniste sort to combine the two concepts. This particular network is often used in the financial industry for making forecasts related to the stock market and as such this Investigation sort to compare the results of stock market forecasts made several implementations of the Wavelet Neural Networks, with different activation functions. The main findings from this investigation found that the ShannonWavelet Neural Network and the GaussianNeural Network were the best performing followed by the Mexican Hat Neural Network.

Researcher: Johannah Moepi, University of the Witwatersrand, Johannesburg
Supervisor: Professor Nukri Komin, University of the Witwatersrand, Johannesburg

With two types of EAS produced, the IACT’s detect both of them including the background events.
As such the IACT’s requires sophisticated image analysis that can be trained to distinguish gamma rays from the hadrons.

Researcher: Tshinanne Mbedzi, University Venda
Supervisor: Professor Turgay Çelik and Dr Michael Klipin, University of the Witwatersrand, Johannesburg

Disease classification may be defined as a category system to which morbid entities are allocated according to the criteria established. The purpose of the ICD is to allow the systematic analysis, interpretation and comparison of data collected at different times on mortality and morbidity[1].
ICD, supported by the WHO(World Health Organization), is translated into a multitude of languages and is the main source of diagnostic definition. [2]. As such the IACT’s requires sophisticated image analysis that can be trained to distinguish gamma rays from the hadrons.

Researcher: Joshua Nel, University of the Witwatersrand, Johannesburg
Supervisor: Professor Rod Alence, University of the Witwatersrand, Johannesburg

The unprecedented growth in size and volume of data requires the examination of novel technologies. The research project applies text mining applications to a section of Truth and Reconciliation Transcripts to determine their viability within investigative journalism.