Showing items 1 - 20 of 22

A survey of machine learning methods applied to anomaly detection on drinking-water quality data

- Dogo, Eustace M., Nwulu, Nnamdi I., Twala, Bhekisipho, Aigbavboa, Clinton

  • Authors: Dogo, Eustace M. , Nwulu, Nnamdi I. , Twala, Bhekisipho , Aigbavboa, Clinton
  • Date: 2019
  • Subjects: Machine learning , Anomaly detection , Deep learning
  • Language: English
  • Type: Article
  • Identifier: http://hdl.handle.net/10210/406768 , uj:34213 , Citation: Dogo, E.M. et al. 2019. A survey of machine learning methods applied to anomaly detection on drinking-water quality data.
  • Description: Abstract: Traditional machine learning (ML) techniques such as support vector machine, logistic regression, and artificial neural network have been applied most frequently in water quality anomaly detection tasks. This paper presents a review of progress and advances made in detecting anomalies in water quality data using ML techniques. The review encompasses both traditional ML and deep learning (DL) approaches. Our findings indicate that: 1) Generally, DL approaches outperform traditional ML techniques in terms of feature learning accuracy and fewer false positive rates. However, is difficult to make a fair comparison between studies because of different datasets, models and parameters employed. 2) We notice that despite advances made and the advantages of the extreme learning machine (ELM), application of ELM is sparsely exploited in this domain. This study also proposes a hybrid DL-ELM framework as a possible solution that could be investigated further and used to detect anomalies in water quality data.
  • Full Text:

Accessing imbalance learning using dynamic selection approach in water quality anomaly detection

- Dogo, Eustace M., Nwulu, Nnamdi I., Twala, Bhekisipho, Aigbavboa, Clinton

  • Authors: Dogo, Eustace M. , Nwulu, Nnamdi I. , Twala, Bhekisipho , Aigbavboa, Clinton
  • Date: 2021
  • Subjects: Classification , Imbalance learning , Dynamic selection
  • Language: English
  • Type: Article
  • Identifier: http://hdl.handle.net/10210/478606 , uj:43260 , Citation: Dogo, E.M.; Nwulu, N.I.; Twala, B.; Aigbavboa, C. Accessing Imbalance Learning Using Dynamic Selection Approach in Water Quality Anomaly Detection. Symmetry 2021, 13, 818. https://doi.org/10.3390/ sym13050818
  • Description: Abstract: Automatic anomaly detection monitoring plays a vital role in water utilities’ distribution systems to reduce the risk posed by unclean water to consumers. One of the major problems with anomaly detection is imbalanced datasets. Dynamic selection techniques combined with ensemble models have proven to be effective for imbalanced datasets classification tasks. In this paper, water quality anomaly detection is formulated as a classification problem in the presences of class imbalance. To tackle this problem, considering the asymmetry dataset distribution between the majority and minority classes, the performance of sixteen previously proposed single and static ensemble classification methods embedded with resampling strategies are first optimised and compared. After that, six dynamic selection techniques, namely, Modified Class Rank (Rank), Local Class Accuracy (LCA), Overall-Local Accuracy (OLA), K-Nearest Oracles Eliminate (KNORA-E), K-Nearest Oracles Union (KNORA-U) and Meta-Learning for Dynamic Ensemble Selection (META-DES) in combination with homogeneous and heterogeneous ensemble models and three SMOTE-based resampling algorithms (SMOTE, SMOTE+ENN and SMOTE+Tomek Links), and one missing data method (missForest) are proposed and evaluated. A binary real-world drinking-water quality anomaly detection dataset is utilised to evaluate the models. The experimental results obtained reveal all the models benefitting from the combined optimisation of both the classifiers and resampling methods. Considering the three performance measures (balanced accuracy, F-score and G-mean), the result also shows that the dynamic classifier selection (DCS) techniques, in particular, the missForest+SMOTE+RANK and missForest+SMOTE+OLA models based on homogeneous ensemble-bagging with decision tree as the base classifier, exhibited better performances in terms of balanced accuracy and G-mean, while the Bg+mF+SMENN+LCA model based on homogeneous ensemble-bagging with random forest has a better overall F1-measure in comparison to the other models.
  • Full Text:

An access optimization approach for FFH-OCDMA system’s fiber bragg gratings encoder

- Nlend, Samuel, Swart, Theo G., Twala, Bhekisipho

Constitutive modelling of INCONEL 718 using artificial neural network

- Abiri, Olufunminiyi, Twala, Bhekisipho

Creating social learning spaces to enhance the learning experience

- Madhav, Natasha, Joseph, Meera K., Twala, Bhekisipho

  • Authors: Madhav, Natasha , Joseph, Meera K. , Twala, Bhekisipho
  • Date: 2015-06-04
  • Subjects: Social media , Educational technology
  • Type: Article
  • Identifier: uj:5145 , http://hdl.handle.net/10210/14187
  • Description: Social media has been used effectively for teaching and learning for many years in developed countries. It seems there needs to be an understanding of the affordances that social media could bring to the learning space in the African context. We use qualitative research and content analysis to see how and why the learners used social media in a private Higher Educational (HE) institution in South Africa to enhance their learning experience. The course was split into face-to-face lectures and social media use by the learners and facilitators. The learners were required to engage with various social media tools to collaborate and share skills with their classmates and facilitators. We also explore the affordance of social media. Content analysis was done to see the participation of the learners in the course’s Facebook page. Data was also collected from the focus groups that led to findings that indicate that WEB 2.0 tools had the potential to support learner collaboration that is self-directed and engaging. Learners adopted social media with confidence and used it to learn beyond the borders of the physical classroom. Daily interactions and the sharing of artefacts resulted in an informal and vibrant learning ecology that became self-sustainable ...
  • Full Text:

Effect of rotor barrier pitch angle on torque ripple production in synchronous reluctance machines

- Muteba, Mbika, Twala, Bhekisipho, Nicolae, Dan-Valentin

Electric power grids distribution generation system for optimal location and sizing — a case study investigation by various optimization algorithms

- Ali, Ahmed, Padmanaban, Sanjeevikumar, Twala, Bhekisipho, Marwala, Tshilidzi

Ensemble missing data techniques for software effort prediction

- Twala, Bhekisipho, Cartwright, Michelle

  • Authors: Twala, Bhekisipho , Cartwright, Michelle
  • Date: 2010
  • Subjects: Software engineering - Forecasting
  • Type: Article
  • Identifier: uj:4670 , ISSN 1088-467X , http://hdl.handle.net/10210/10428
  • Description: Constructing an accurate effort prediction model is a challenge in software engineering. The development and validation of models that are used for prediction tasks require good quality data. Unfortunately, software engineering datasets tend to suffer from the incompleteness which could result to inaccurate decision making and project management and implementation. Recently, the use of machine learning algorithms has proven to be of great practical value in solving a variety of software engineering problems including software prediction, including the use of ensemble (combining) classifiers. Research indicates that ensemble individual classifiers lead to a significant improvement in classification performance by having them vote for the most popular class. This paper proposes a method for improving software effort prediction accuracy produced by a decision tree learning algorithm and by generating the ensemble using two imputation methods as elements. Benchmarking results on ten industrial datasets show that the proposed ensemble strategy has the potential to improve prediction accuracy compared to an individual imputation method, especially if multiple imputation is a component of the ensemble.
  • Full Text:

First principle leakage current reduction technique for CMOS devices

- Tsague, Hippolyte Djonon, Twala, Bhekisipho

Forecasting electricity consumption in South Africa : ARMA, neural networks and neuro-fuzzy systems

- Marwala, Lufuno, Twala, Bhekisipho

  • Authors: Marwala, Lufuno , Twala, Bhekisipho
  • Date: 2014
  • Subjects: Neural networks , Electric power consumption , Fuzzy systems , ARMA models
  • Type: Article
  • Identifier: uj:5090 , http://hdl.handle.net/10210/13678
  • Description: This paper presents an experiment that consists of constructing auto-regressive moving average (ARMA), neural networks and neuro-fuzzy models with historical electricity consumptiontime series data to create models that can be used to forecastconsumption inthe future. The data was sampled on a monthly basis from January 1985 to December 2011. An ARMA,multilayer perceptron neural network with back propagation and neuro-fuzzy modelling technique which combines Takagi-Sugeno fuzzy models and neural networks were used to create the models for one step ahead forecasting. The results of the three techniques were compared and the results show that neurofuzzy models outperformed the neural network and ARMA models in terms of accuracy.
  • Full Text:

Gold mine dam levels and energy consumption classification using artificial intelligence methods

- Hasan, Ali N., Twala, Bhekisipho, Marwala, Tshilidzi

Improving single classifiers prediction accuracy for underground water pump station in a gold mine using ensemble techniques

- Hasan, Ali N, Twala, Bhekisipho

  • Authors: Hasan, Ali N , Twala, Bhekisipho
  • Date: 2015
  • Subjects: Mutual information , Support vector machines , Prediction
  • Language: English
  • Type: Article
  • Identifier: http://hdl.handle.net/10210/85649 , uj:19373 , Citation: Hasan,, A.N. & Twala, B. 2015. Improving single classifiers prediction accuracy for underground water pump station in a gold mine using ensemble techniques.
  • Description: Abstract: In this paper six single classifiers (support vector machine, artificial neural network, naïve Bayesian classifier, decision trees, radial basis function and k nearest neighbors) were utilized to predict water dam levels in a deep gold mine underground pump station. Also, Bagging and Boosting ensemble techniques were used to increase the prediction accuracy of the single classifiers. In order to enhance the prediction accuracy even more a mutual information ensemble approach is introduced to improve the single classifiers and the Bagging and Boosting prediction results. This ensemble is used to classify, thus monitoring and predicting the underground water dam levels on a single-pump station deep gold mine in South Africa, Mutual information theory is used in order to determine the classifiers optimum number to build the most accurate ensemble. In terms of prediction accuracy, the results show that the mutual information ensemble over performed the other used ensembles and single classifiers and is more efficient for classification of underground water dam levels. However the ensemble construction is more complicated than the Bagging and Boosting techniques.
  • Full Text:

Improving the performance of the Rpper in insurance risk classification : a comparative study using feature selection

- Duma, Mlungisi, Twala, Bhekisipho, Marwala, Tshilidzi

  • Authors: Duma, Mlungisi , Twala, Bhekisipho , Marwala, Tshilidzi
  • Date: 2011
  • Subjects: Automatic relevance determination , Ripper algorithm , Artificial neural networks
  • Type: Article
  • Identifier: uj:4709 , http://hdl.handle.net/10210/10954
  • Description: The Ripper algorithm is designed to generate rule sets for large datasets with many features. However, it was shown that the algorithm struggles with classification performance in the presence of missing data. The algorithm struggles to classify instances when the quality of the data deteriorates as a result of increasing missing data. In this paper, a feature selection technique is used to help improve the classification performance of the Ripper model. Principal component analysis and evidence automatic relevance determination techniques are used to improve the performance. A comparison is done to see which technique helps the algorithm improve the most. Training datasets with completely observable data were used to construct the model and testing datasets with missing values were used for measuring accuracy. The results showed that principal component analysis is a better feature selection for the Ripper in improving the classification performance.
  • Full Text:

Leakage current minimisation and power reduction techniques using sub-threshold design

- Tsague, Hippolyte Djonon, Twala, Bhekisipho

Modelling the flow stress of alloy 316L using a multi-layered feedforward neural network with Bayesian regularization

- Abiri, Olufunminiyi, Twala, Bhekisipho

Optimal placement model of TCSC in power system network considering the budget available

- Malatji, Esrom Mahlatsi, Twala, Bhekisipho, Mbuli, Nhlanhla

Optimization of the compressed air-usage in South African mines

- Hassan, Ali, Ouahada, Khmaies, Marwala, Tshilidzi, Twala, Bhekisipho

Predicting engineering student success using machine learning

- Taodzera, Tatenda, Twala, Bhekisipho, Carroll, Johnson

  • Authors: Taodzera, Tatenda , Twala, Bhekisipho , Carroll, Johnson
  • Date: 2017
  • Language: English
  • Type: Conference proceedings
  • Identifier: http://hdl.handle.net/10210/236143 , uj:24160 , Citation: Taodzera, T., Twala, B. & Carroll, J. 2017. Predicting engineering student success using machine learning.
  • Description: Abstract: Recent years have seen an increase in the number of students from diverse backgrounds enrolling into South African universities, presenting many challenges. Some students struggle with their academic choices, and universities struggle to understand and address the individual needs of such a diverse student base. Fortunately, vast amounts of student information have been collected and stored, giving an opportunity for researchers in educational data mining to derive some useful insights from this data to help both the universities and students. This research aims to identify factors that contribute to the success and or failure of a student, then predict the future performance of the student at enrolment. By using data pre-processing techniques, the experiments identify the most significant success factors from the data at enrolment time. The most significant factors can then be used to identify students who may need extra support, and the nature of those factors can help determine the manner of support needed. This study implemented and evaluated the effectiveness of the most commonly used and new machine learning algorithms in predicting student performance on a sample of 1366 engineering students. The results show various degrees of success in predicting student performance, and it is hoped that these findings will guide the selection of machine learning algorithms for future studies.
  • Full Text:

Predicting mine dam levels and energy consumption using artificial intelligence methods

- Hasan, Ali N., Twala, Bhekisipho, Marwala, Tshilidzi

Predicting software faults in large space systems using machine learning techniques

- Twala, Bhekisipho

  • Authors: Twala, Bhekisipho
  • Date: 2011-07-04
  • Subjects: Software metrics , Machine learning , Fault-proneness prediction , Ensemble classifiers
  • Type: Article
  • Identifier: uj:5314 , ISSN 0011-748X , http://hdl.handle.net/10210/7754
  • Description: Recently, the use of machine learning (ML) algorithms has proven to be of great practical value in solving a variety of engineering problems including the prediction of failure, fault, and defect-proneness as the space system software becomes complex. One of the most active areas of recent research in ML has been the use of ensemble classifiers. How ML techniques (or classifiers) could be used to predict software faults in space systems, including many aerospace systems is shown, and further use ensemble individual classifiers by having them vote for the most popular class to improve system software fault-proneness prediction. Benchmarking results on four NASA public datasets show the Naive Bayes classifier as more robust software fault prediction while most ensembles with a decision tree classifier as one of its components achieve higher accuracy rates.
  • Full Text:
  • Disclaimer
  • Privacy
  • Copyright
  • Contact