Electrical resistivity tomography (ERT) is a geophysical technique for modelling the properties of the shallow subsurface. The technique provides a powerful tool for a volumetric representation of the spatial properties and spatio-temporal systems below the ground by indirectly measuring electrical properties. ERT has wide-reaching applications for imaging and monitoring in fields such as mineral exploration, infrastructure, and groundwater modelling. Developing tools that can perform predictions and analysis on the resistivity models with limited intervention will allow for ERT systems to be deployed remotely so that they might serve as an alert system, for example, in areas at risk of landslides, or groundwater contamination. However, the nature of indirect observation in ERT imaging means that there is a high degree of uncertainty in the resolved models, resulting from systematic artefacts that occur in inversion processes and from the fact that the underlying structures and processes cannot be directly observed. This thesis presents a number of developments in automating the analysis and prediction of directly and indirectly observed uncertain systems, both static and dynamic. Drawing from principles in both fuzzy logic and probability, particularly Bayesian statistics, the different representations of uncertainty are exploited and utilised to make meaningful estimates of properties and parameters in noisy systems. The key contributions of the research presented include the unique combination of fuzzy inference systems in a recursive Bayesian estimator to resolve systems under the influence of multiple uncertain dynamic processes. Furthermore, frameworks for robustly isolating features with quantified certainty and for automatically tracking tracer moments in hydrodynamic systems are proposed and applied to a number of real-world case studies.

Considering the multiobjective nature of real-world optimisation problems requiring a search for optimal trade-off solutions, many multiobjective metaheuristics have been proposed in the scientific literature. As observed in previous studies, different approaches show strengths on different problems. A research question would be how to combine the strengths of those multiple multiobjective approaches to obtain improved performance across a range of problems. Hyper-heuristics, which emerged as general-purposed search methods with reusable components, are one of the design philosophies to achieve that goal. Hyper heuristics perform search over the space of (meta)heuristics by either selecting an appropriate (meta)heuristic or generating a new one from given components. Hierarchically, the (meta)heuristics are called low-level (meta)heuristics since they work underneath the selection or generation strategies of the hyper-heuristics. This feature leads to increased generality level for search methods and enables multiobjective hyper-heuristics to be applied to a wide range of problem domains than a metaheuristic tailored for a particular problem domain. A crucial component in such hyper-heuristics is learning and hence, various online and offline learning mechanisms have been adopted within hyper-heuristics. In this thesis, the focus is online learning based selection hyper-heuristics for multiobjective optimisation. To gain insights of the behaviour and roles of online learning mechanisms played in selection hyper-heuristics, nine hyper-heuristics including online learning based, predefined sequence based and random choice based are applied to and analysed on an 'unseen' real-world problem, wind farm layout optimisation. The empirical results show that selection hyper-heuristics can indeed exploit the strengths of different MOEAs. Meanwhile, it also suggests two research directions: find a reasonable combination of low-level multiobjective evolutionary algorithms (MOEAs) for the selection hyper-heuristic framework to perform search on, and come up with a more effective online learning mechanism for the hyper-heuristic framework to exploit the strengths of different low-level MOEAs. Therefore, a critical review of different types of MOEAs is carried out in order to develop a better understanding of their nature, advantages and disadvantages. This review would lead to a more informed decision on the choice of the low-level metaheuristic set that selection hyper-heuristics can operate on. In addition, based on the investigation of hypervolume guided EAs, an improved version of such algorithm is proposed in this thesis, which later is used as one of the low-level MOEAs in the proposed selection hyper-heuristics. Following this, two learning automata based selection hyper-heuristics for multiobjective optimisation are proposed which select an appropriate metaheuristic to perform at a given time based on the information gathered during the search. Due to the complicated nature of multiobjective optimisation, the learning automata in the proposed hyper-heuristics is employed in a non-traditional way and novel components are also designed for making the best use of the learned information. The proposed hyper-heuristics are compared with a range of multiobjective approaches including a state-of-the-art online learning based selection hyper-heuristic on four problem domains including two mathematical benchmark functions and two real-world problems. The experimental results demonstrate the superior performance and generality of the proposed approach. To further challenge the proposed hyper-heuristics, different numbers and types of metaheuristics are incorporated as the low-level metaheuristics and combined with different acceptance strategies. The proposed learning automata based hyper-heuristics are the best-performed ones based on the performance indicator, hypervolume µ-norm.

The need for teachers of Elementary School children to learn to program or rather to understand the Computational Thinking behind programming has been accelerated in many countries by the mandated teaching of programming in the Elementary School context. Many steps have been taken in order to create awareness of this issue, such as the Computing At Schools initiative (CAS) which is established in the UK. CAS aims to support teaching in computing and connected fields in UK schools. Moreover, in the USA the Computer Science Teachers Association (CSTA) was established to meet the purpose of informing and advising about the current development of computational thinking and to investigate and disseminate teaching and learning resources related to computational thinking. In Singapore research has been conducted by the government agency Infocomm Development Authority of Singapore (IDA) whereby the major goal is to meet the needs in the ICT sector and ultimately to focus and inspire learners about programming. The research for this thesis involves the development of a training scheme for pre-service teachers that will introduce them to computational thinking through the use of the Flash Action Script Development environment. Flash Action Scripts - amongst several other tools - are used as a tool for creating interactive content and because Flash is one of the premiere tools used to create content for the internet; a tool programmed with Flash looks practically the same in every browser and on every operating system. Flash Action scripts use traditional coding skills but permit the user to see how each piece of code affects the running or execution of the program, allowing the user to have an instant visual understanding of what the code is doing. It is also widely available within university campuses. A major problem in promoting the teaching of programming and computational thinking to Elementary School teachers is that the majority of such teachers have no concept of how to program and naturally are not motivated to learn programming. Experienced teachers involved in the current study felt that programming was too complicated and thus it was hard to gain fluency in programming. Student teachers who had no previous experience in programming were, however, easier to get engaged in learning programming principles. Eighty percent of this group found Action Scripting a useful tool to understand basic programming and scripting. The need to teach programming will motivate most but to learn through a tool that can be seen to have intrinsic value in their role as teachers has a greater potential of success. This thesis defines the design and implementation of a tool to use the learning of Flash Action Scripting as a motivational mechanism for pre-service teachers. The intrinsic value to them is intended to be utilisation of the learned Action Scripting skills to produce their own teaching material. Initial results indicate an enhanced engagement and motivation to learn to program and improved confidence in doing so. As projected the pre-service teachers had a more positive attitude towards the potential of the learning tool but both they and the in-service teachers had improved attitudes and enthusiasm after the experiment. The results show that both pre-service and in-service teachers can be trained to be designers and producers of digital courseware in the previous absence of computational thinking skills and definitely they can acquire skills in computer programming such as Flash Action Scripts.

Monads and applicative functors are staple design patterns to handle effects in pure functional programming, especially in Haskell with its built-in syntactic support. Over the last decade, however, practical needs and theoretical research have given rise to generalisations of monads and applicative functors. Examples are graded, indexed and constrained monads. The problem with these generalisations is that no unified representation of standard and generalised monads or applicatives exists in theory or practice. As a result, in Haskell, each generalisation has its own representation and library of functions. Hence, interoperability among the different notions is hampered and code is duplicated. To solve the above issues, I first survey the three most wide-spread generalisations of monads and applicatives: their graded, indexed and constrained variations. I then examine two approaches to give them a unified representation in Haskell: polymonads and supermonads. Both approaches are embodied in plugins for the Haskell compiler GHC that address most of the identified concerns. Finally, I examine category theory and propose unifying categorical models that encompass the three discussed generalisations together with the standard notions of monad and applicative.

This thesis presents a detailed study about one of the Multiple Criteria Decision Making (MCDM) models, namely Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), based on fuzzy set theory (FST) by focusing on improving modelling uncertain information provided by a group of decision makers (DMs). An exploration of issues and limitations in current models of standard TOPSIS and fuzzy TOPSIS were made. Despite many variations of type-1 fuzzy TOPSIS (T1-TOPSIS) model, none of the studies explaining the details of the key stages of standard TOPSIS (non-fuzzy) and T1-TOPSIS are based on a step-wise procedure. A detailed study was conducted which involve the process of identifying the limitations of standard TOPSIS and T1-TOPSIS. Based on this, a novel contribution on the comparison between these two models in systematic stepwise procedure was given. This study successfully identified and discussed the limitations, issues and challenges which have not been investigated sufficiently in the context of T1-TOPSIS model. Based on this exploration, further investigation of multiple variants of the extension of the fuzzy TOPSIS model for solving the MCDM problem was made with the primary aim of detailing the steps involved. One challenge that has risen is that it is not straightforward to differentiate between the multiple variants of TOPSIS existing today. A systematic comparison was made between standard T1-TOPSIS model with the recently extended model to show the differences between both models and to provide context for their respective strengths and limitations both in the complexity of application and expressiveness of results. Based on the resulting comparison, the differences in the steps implemented by these two Fuzzy TOPSIS models were highlighted throughout the worked example. Furthermore, this task highlights the ability of both models to handle different levels of uncertainty. Following the exploration of issues and limitations of the current model, as well as a comparative study, a novel extension of type-2 fuzzy TOPSIS model is proposed in this thesis which suggests providing an interval-valued output to reflect the uncertainties and to model subjective information. The proposed model enables to uniquely captures input uncertainty (i.e., decision-makers' preferences) in the decision-making outputs and provide a direct mapping of uncertainty in the inputs to outputs. By keeping the output values in interval form, the proposed model reduces the loss of information and maximises the potential benefit of using Interval Type-2 Fuzzy Sets (IT2 FSs). To demonstrate the MCDM problems when a various level of uncertainty is introduced, a novel experimental method was proposed in this study. The primary aim is to explore the use of IT2 FSs in handling uncertainty based on the TOPSIS model. This experiment was conducted to show how the variation of uncertainty levels in the input affects the final outputs. An implementation of the proposed model to two different case studies was conducted to evaluate the proposed model. The proposed model for the first time generates an interval-valued output. As intervals can, for example, exhibit partial overlap, a novel extended measure is proposed to compare the resulting interval-valued output from various cases (i.e., overlapping and non-overlapping) of the interval with considering uncertainty.

The root system of a plant is responsible for supplying it with essential nutrients. The plant's ability to explore the surrounding soil is largely determined by its root system architecture (RSA), which varies with both genetic and environmental conditions. X-ray micro computed tomography (µCT) is a powerful tool allowing the non-invasive study of the root system architecture of plants grown in natural soil environments, providing both 3D descriptions of root architecture and the ability to make multiple measurements over a period of time. Once volumetric µCT data is acquired, the root system must first be segmented from the surrounding soil environment and then described. Automated and semi-automated software tools can be used to extract roots from µCT images, but current methods for the recovery of RSA traits from the resulting volumetric descriptions are somewhat limited. This thesis presents a novel tool (RooTh) which, given a segmented µCT image, skeletonises the root system and quantifies global and local root traits with minimal user interaction. The computationally inexpensive method used takes advantage of curve-fitting and active contours to find the optimal skeleton and thus evaluate root traits objectively. A small-scale experiment was conducted to validate and compare root traits extracted using the method presented here alongside other 2D imaging tools. The results show a good degree of correlation between the two methods.

Currently, fossil fuels contribute a large number of high value chemicals that are used on a daily basis. Crude oil is cracked to give a number of high value chemicals, including vehicle fuels as well as chemicals and solvents that are used daily both commercially and industrially. However, fossil fuel reserves are in decline, with research going into alternatives to obtain these useful chemicals, one of which is biofuels. Biofuels can be generated in a number of ways, one of which is the fermentation of acetogenic bacteria, microorganisms that generate acetate as a product of anaerobic metabolism. Clostridium autoethanogenum is an acetogen that can grow on one carbon gases as its feedstock, and can be used to generate valuable chemicals, with scope to develop the range of metabolic products further. One aim of this project was to investigate the metabolic flux through pathways of the bacterium using isotopically labelled compounds, which would be assessed by mass spectrometry and NMR. Following on from this, design of inhibitors for the enzymes of the pathways with a view to drive the metabolic processes towards higher value chemical compounds by ‘switching off’ other branches of the pathway. Putative small molecule mimics of acetyl-CoA, SNAC thioesters, were synthesised and tested for uptake and activity in whole cell growth experiments with C. autoethanogenum, and determined to be unsuccessful. Further to this, compounds were designed and synthesised to replace pantothenic acid in the growth media, which were not tested in growth experiments. A library of inhibitor compounds was synthesised and tested against recombinantly purified acetate kinase. A number of compounds were shown to inhibit the enzyme, and the mode of inhibition was determined, as well as IC50 and Ki values for each. This project operated as part of a larger GASCHEM project in the Synthetic Biology Research Centre at the University of Nottingham.

Supervised k nearest neighbour and unsupervised hierarchical agglomerative clustering algorithm can be enhanced through word mover’s distance-based sentence distance metric to offer superior context-aware sentence categorisation performance. Advanced neural network-oriented classifier is able to achieve competing result on the benchmark streams via an aggregated recurrent unit incorporated with sophis- ticated convolving layer. The continually increasing number of textual snippets produced each year ne- cessitates ever improving information processing methods for searching, retrieving, and organising text. Central to these information processing methods are sentence classification and clustering, which have become an important application for nat- ural language processing and information retrieval. This present work proposes three novel sentence categorisation frameworks, namely hierarchical agglomerative clustering-word mover’s distance, k nearest neighbour-word mover’s distance, and convolutional recurrent neural network. Hierarchical agglomerative clustering-word mover’s distance employs word mover’s distance distortion function to effectively cluster unlabelled sentences into nearby centroid. K nearest neighbour-word mover’s distance classifies testing textual snippets through word mover’s distance-based sen- tence similarity. Both models are from the spectrum of count-based framework since they apply term frequency statistics when building the vector space matrix. Experimental evaluation on the two unsupervised learning data-sets show better per- formance of hierarchical agglomerative clustering-word mover’s distance over other competitors on mean squared error, completeness score, homogeneity score, and v-measure value. For k nearest neighbour-word mover’s distance, two benchmark textual streams are experimented to verify its superior classification performance against comparison algorithms on precision rate, recall ratio, and F1 score. Per- formance comparison is statistically validated via Mann-Whitney-U test. Through extensive experiments and results analysis, each research hypothesis is successfully verified to be yes. Unlike traditional singleton neural network, convolutional recurrent neural net- work model incorporates character-level convolutional network with character-aware recurrent neural network to form a combined framework. The proposed model ben- efits from character-aware convolutional neural network in that only salient features are selected and fed into the integrated character-aware recurrent neural network. Character-aware recurrent neural network effectively learns long sequence semantics via sophisticated update mechanism. The experiment presented in current thesis compares convolutional recurrent neural network framework against the state-of- the-art text classification algorithms on four popular benchmarking corpus. The present work also analyses three different recurrent neural network hidden recurrent cells’ impact on performance and their runtime efficiency. It is observed that min- imal gated unit achieves the optimal runtime and comparable performance against gated recurrent unit and long short-term memory. For term frequency-inverse docu- ment frequency-based algorithms, the current experiment examines word2vec, global vectors for word representation, and sent2vec embeddings and reports their perfor- mance differences. Performance comparison is statistically validated through Mann- Whitney-U test and the corresponding hypotheses are tested to be yes through the reported statistical analysis.

A hyper-heuristic is a search method or learning mechanism for selecting or generating heuristics to solve computational search problems. Researchers classify hyper-heuristics according to the source of feedback during learning: Online learning hyper-heuristics learn while solving a given instance of a problem; Offline learning hyper-heuristics learn from a set of training instances, a method that can generalise to unseen instances. Genetic programming (GP) can be considered a specialization of the more widely known genetic algorithms (GAs) where each individual is a computer program. GP automatically generates computer programs to solve specified tasks. It is a method of searching a space of computer programs. GP can be used as a kind of hyper-heuristic to be a learning algorithm when it uses some feedback from the search process. Our research mainly uses genetic programming as offline hyper-heuristic approach to automatically design various heuristics for evolutionary programming.

This thesis is composed of three separate parts. The first part deals with definability and productivity issues of equational systems defining polymorphic stream functions. The main result consists of showing such systems composed of only unary stream functions complete with respect to specifying computable unary polymorphic stream functions. The second part deals with syntactic and semantic notions of isomorphism of finitary inductive types and associated decidability issues. We show isomorphism of so-called guarded types decidable in the set and syntactic model, verifying that the answers coincide. The third part deals with homotopy levels of hierarchical univalent universes in homotopy type theory, showing that the n-th universe of n-types has truncation level strictly n+1.