Nelishia Pillay

Prof. Nelishia Pillay

University of Pretoria
orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 July 2022 in Plants
Verified
Journal Article
Open access
PDF file
HCHamish Andrew Craze
Hamish Andrew Craze
NPNelishia Pillay
Nelishia Pillay
FJFourie Joubert
Fourie Joubert
DBDave Kenneth Berger
Dave Kenneth Berger
Maize yields worldwide are limited by foliar diseases that could be fungal, oomycete, bacterial, or viral in origin. Correct disease identification is critical for farmers to apply the correct control measures, such as fungicide sprays. Deep learning has the potential for automated disease classification from images of leaf symptoms. We aimed to develop a classifier to identify gray leaf spot (GLS) disease of maize in field images where mixed diseases were present (18,656 images after augmentation). In this study, we compare deep learning models trained on mixed disease field images with and without background subtraction. Performance was compared with models trained on PlantVillage images with single diseases and uniform backgrounds. First, we developed a modified VGG16 network referred to as “GLS_net” to perform binary classification of GLS, which achieved a 73.4% accuracy. Second, we used MaskRCNN to dynamically segment leaves from backgrounds in combination with GLS_net to identify GLS, resulting in a 72.6% accuracy. Models trained on PlantVillage images were 94.1% accurate at GLS classification with the PlantVillage testing set but performed poorly with the field image dataset (55.1% accuracy). In contrast, the GLS_net model was 78% accurate on the PlantVillage testing set. We conclude that deep learning models trained with realistic mixed disease field data obtain superior degrees of generalizability and external validity when compared to models trained using idealized datasets.

17 November 2020 in Mathematics
Verified
Journal Article
Open access
PDF file
Jorge Mario Cruz-DuarteJorge Mario Cruz-Duarte
Jorge Mario Cruz-Duarte
José Carlos Ortiz-BaylissJosé Carlos Ortiz-Bayliss
José Carlos Ortiz-Bayliss
Ivan AmayaIvan Amaya
Ivan Amaya
YSYong Shi
Yong Shi
Hugo Terashima-MarinHugo Terashima-Marin
Hugo Terashima-Marin
NPNelishia Pillay
Nelishia Pillay
Metaheuristics have become a widely used approach for solving a variety of practical problems. The literature is full of diverse metaheuristics based on outstanding ideas and with proven excellent capabilities. Nonetheless, oftentimes metaheuristics claim novelty when they are just recombining elements from other methods. Hence, the need for a standard metaheuristic model is vital to stop the current frenetic tendency of proposing methods chiefly based on their inspirational source. This work introduces a first step to a generalised and mathematically formal metaheuristic model, which can be used for studying and improving them. This model is based on a scheme of simple heuristics, which perform as building blocks that can be modified depending on the application. For this purpose, we define and detail all components and concepts of a metaheuristic (i.e., its search operators), such as heuristics. Furthermore, we also provide some ideas to take into account for exploring other search operator configurations in the future. To illustrate the proposed model, we analyse search operators from four well-known metaheuristics employed in continuous optimisation problems as a proof-of-concept. From them, we derive 20 different approaches and use them for solving some benchmark functions with different landscapes. Data show the remarkable capability of our methodology for building metaheuristics and detecting which operator to choose depending on the problem to solve. Moreover, we outline and discuss several future extensions of this model to various problem and solver domains.

7 August 2020 in Journal of the Operational Research Society
Verified
Journal Article
Closed access
NPNelishia Pillay
Nelishia Pillay
Rong QuRong Qu
Rong Qu
Limited attention has been paid to assessing the generality performance of hyper-heuristics. The performance of hyper-heuristics has been predominately assessed in terms of optimality which is not ideal as the aim of hyper-heuristics is not to be competitive with state of the art approaches but rather to raise the level of generality, i.e. the ability of a technique to produce good results for different problem instances or problems rather than the best results for some instances and poor results for others. Furthermore from existing literature in this area it is evident that different hyper-heuristics aim to achieve different levels of generality and need to be assessed as such. To cater for this the paper firstly presents a new taxonomy of four different levels of generality that can be attained by a hyper-heuristic based on a survey of the literature. The paper then proposes a performance measure to assess the performance of different types of hyper-heuristics at the four levels of generality in terms of generality rather than optimality. Three case studies from the literature are used to demonstrate the application of the generality performance measure. The paper concludes by examining how the generality measure can be combined with measures of other performance criteria, such as optimality, to assess hyper-heuristic performance on more than one criterion.

21 February 2024 in Genetic Programming and Evolvable Machines
Verified
Journal Article
Open access
PDF file
KPKyle Pretorius
Kyle Pretorius
NPNelishia Pillay
Nelishia Pillay
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.

18 June 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Jorge Mario Cruz-DuarteJorge Mario Cruz-Duarte
Jorge Mario Cruz-Duarte
José Carlos Ortiz-BaylissJosé Carlos Ortiz-Bayliss
José Carlos Ortiz-Bayliss
Ivan AmayaIvan Amaya
Ivan Amaya
NPNelishia Pillay
Nelishia Pillay
Optimisation has been with us since before the first humans opened their eyes to natural phenomena that inspire technological progress. Nowadays, it is quite hard to find a solver from the overpopulation of metaheuristics that properly deals with a given problem. This is even considered an additional problem. In this work, we propose a heuristic-based solver model for continuous optimisation problems by extending the existing concepts present in the literature. We name such solvers ‘unfolded’ metaheuristics (uMHs) since they comprise a heterogeneous sequence of simple heuristics obtained from delegating the control operator in the standard metaheuristic scheme to a high-level strategy. Therefore, we tackle the Metaheuristic Composition Optimisation Problem by tailoring a particular uMH that deals with a specific application. We prove the feasibility of this model via a two-fold experiment employing several continuous optimisation problems and a collection of diverse population-based operators with fixed dimensions from ten well-known metaheuristics in the literature. As a high-level strategy, we utilised a hyper-heuristic based on Simulated Annealing. Results demonstrate that our proposed approach represents a very reliable alternative with a low computational cost for tackling continuous optimisation problems with a tailored metaheuristic using a set of agents. We also study the implication of several parameters involved in the uMH model and their influence over the solver performance.

1 January 2022
Verified
Book Chapter
Closed access
PDF file
TGTapotosh Ghosh
Tapotosh Ghosh
HBHasan Al Banna
Hasan Al Banna
JNJaber Al Nahian
Jaber Al Nahian
Mohammed Shamim KaiserMohammed Shamim Kaiser
Mohammed Shamim Kaiser
Mufti MahmudMufti Mahmud
Mufti Mahmud
NPNelishia Pillay
Nelishia Pillay