Methods of construction of OWA operators from data

This paper investigates the problem of obtaining the weights of the ordered weighted aggregation (OWA) operators from observations. The problem is formulated as a restricted least squares and uniform approximation problems. We take full advantage of the linearity of the problem. In the former case, a well known technique of non-negative least squares is used. In a case of uniform approximation, we employ a recently developed cutting angle method of global optimisation. Both presented methods give results superior to earlier approaches, and do not require complicated nonlinear constructions. Additional restrictions, such as degree of orness of the operator, can be easily introduced

Extending robust industrial optimisation using simulation and reinforcement learning

Traditional optimisation methods are incapable of capturing the complexity of today's dynamic manufacturing systems. A new methodology, integrating simulation models and intelligent learning agents, was successfully applied to identify solutions to a fundamental scheduling problem. The robustness of this approach was then demonstrated through a series of real-world industrial applications.

A systems approach to shape and topology optimisation of mechanical structures

Optimisation techniques have become more and more important as the possibility of simulating complex mechanical structures has become a reality. A common tool in the layout design of structural parts is the topology optimisation method, which finds an optimum material distribution within a given geometrical design space to best meet loading conditions and constraints. Another important method is shape optimisation, which optimises weight given parametric geometric constraints. In the case of complex shaped parts or elaborate assemblies, for example automobile body structures, shape optimisation is still hard to do; mainly due to the difficulty in translating shape design parameters into meaningful analysis models. Tools like the parametric geometry package SFE CONCEPT are designed to mitigate these issues. Nevertheless, shape methods usually cannot suggest new load path configurations, while topology methods are often confined to single parts. To overcome these limitations the authors have developed a method that combines both approaches into an Integral Shape/Topology Method (IST) that is capable of finding new optimal solutions. This is achieved by an automated optimisation loop and can be applied for both thin walled structures as well as solid 3D geometries. When optimising structures by applying IST, global optimum solutions can be determined that may not be obtained with isolated shape- or topology-optimisation methods.

An integrated approach to shape and topology optimisation of mechanical structures

In mechanical engineering, simulation and optimisation methods have become indispensable. The thesis looks into a novel way to combine shape and topology optimisation approaches. The proposed method - named IST for Integrated Shape And Topology Optimisation - proves to be beneficial for many application in the automotive and aerospace industry.

Subsuming methods: finding new optimisation opportunities in object-oriented software

The majority of existing application profiling techniques ag- gregate and report performance costs by method or call- ing context. Modern large-scale object-oriented applications consist of thousands of methods with complex calling pat- terns. Consequently, when profiled, their performance costs tend to be thinly distributed across many thousands of loca- tions with few easily identifiable optimisation opportunities. However experienced performance engineers know that there are repeated patterns of method calls in the execution of an application that are induced by the libraries, design patterns and coding idioms used in the software. Automati- cally identifying and aggregating costs over these patterns of method calls allows us to identify opportunities to improve performance based on optimising these patterns. We have developed an analysis technique that is able to identify the entry point methods, which we call subsuming methods, of such patterns. Our ofiine analysis runs over previously collected runtime performance data structured in a calling context tree, such as produced by a large number of existing commercial and open source profilers. We have evaluated our approach on the DaCapo bench- mark suite, showing that our analysis significantly reduces the size and complexity of the runtime performance data set, facilitating its comprehension and interpretation. We also demonstrate, with a collection of case studies, that our analysis identifies new optimisation opportunities that can lead to significant performance improvements (from 20% to over 50% improvement in our case studies).

Towards insightful algorithm selection for optimisation using meta-learning concepts

In this paper we propose a meta-learning inspired framework for analysing the performance of meta-heuristics for optimization problems, and developing insights into the relationships between search space characteristics of the problem instances and algorithm performance. Preliminary results based on several meta-heuristics for well-known instances of the Quadratic Assignment Problem are presented to illustrate the approach using both supervised and unsupervised learning methods.

The optimisation of ultrasonic cleaning procedures for dairy fouled ultrafiltration membranes

Ultrafiltration (UF) of whey is a major membrane based process in the dairy industry. However, commercialization of this application has been limited by membrane fouling, which has a detrimental influence on the permeation rate. There are a number of different chemical and physical cleaning methods currently used for cleaning a fouled membrane. It has been suggested that the cleaning frequency and the severity of such cleaning procedures control the membrane lifetime. The development of an optimal cleaning strategy should therefore have a direct implication on the process economics. Recently, the use of ultrasound has attracted considerable interest as an alternative approach to the conventional methods. In the present study, we have studied the ultrasonic cleaning of polysulfone ultrafiltration membranes fouled with dairy whey solutions. The effects of a number of cleaning process parameters have been examined in the presence of ultrasound and results compared with the conventional operation. Experiments were conducted using a small single sheet membrane unit that was immersed totally within an ultrasonic bath. Results show that ultrasonic cleaning improves the cleaning efficiency under all experimental conditions. The ultrasonic effect is more significant in the absence of surfactant, but is less influenced by temperature and transmembrane pressure. Our results suggest that the ultrasonic energy acts primarily by increasing the turbulence within the cleaning solution.

Construction and optimisation of prediction intervals for neural network models

This thesis makes contributions to basic and fundamental research in the field of prediction interval construction using neural network models. It proposes novel methods for objective assessment, rapid construction, and optimisation of neural network-based prediction intervals for uncertainty quantification.

The whole-body fatty acid balance method : examples of its potential for feed efficiency and product quality optimisation in fish and poultry

Dietary lipids and fatty acids are not only fundamental in determining animal performance, but also determine the eating qualities of animal products. Several methods have been used to quantify fatty acid metabolism but most involve expensive in vitro approaches that are not suitable for most laboratories. Furthermore, there is considerable variation between methods with regard to enzyme activity, which makes comparison of results between studies difficult. The recently developed whole-body fatty acid balance method (WBFABM) is a simple and reliable in vivo method for assessing fatty acid metabolism, including rates of liponeogenesis and de novo fatty acid production, β-oxidation of fatty acids and bioconversion (elongation and desaturation) of fatty acids to long-chain polyunsaturated fatty acids. Initially developed for implementation with a fish model, the WBFABM has proven to be a simple and effective method that can be used in any laboratory equipped with a gas chromatography unit. Since its development, it has been used in several farmed finfish feeding trials and in broiler chicken feeding trials. The WBFABM is currently used at research institutions worldwide and its use is increasing in popularity among animal scientists. With this method, it is possible to track the fate of individual dietary fatty acids within the body. The WBFABM could contribute significantly to information generated by animal feeding trials.

Detecting K-complexes for sleep stage identification using nonsmooth optimisation

The process of sleep stage identification is a labour-intensive task that involves the specialized interpretation of the polysomnographic signals captured from a patient’s overnight sleep session. Automating this task has proven to be challenging for data mining algorithms because of noise, complexity and the extreme size of data. In this paper we apply nonsmooth optimization to extract key features that lead to better accuracy. We develop a specific procedure for identifying K-complexes, a special type of brain wave crucial for distinguishing sleep stages. The procedure contains two steps. We first extract “easily classified” K-complexes, and then apply nonsmooth optimization methods to extract features from the remaining data and refine the results from the first step. Numerical experiments show that this procedure is efficient for detecting K-complexes. It is also found that most classification methods perform significantly better on the extracted features.

A modelling framework for optimisation of commodity production by minimising the impact of climate change

The agricultural sector is vulnerable to the impact of climate change due to decreasing rainfall, increasing temperature, and the frequency of extreme weather events. A modelling framework was developed and applied to identify issues, problems and opportunities arising in regional agricultural systems as a consequence of climate change. This integrated framework blends together land suitability analysis, uncertainty analysis and an optimisation approach to establish optimal agricultural land-use patterns on a regional scale for current and possible future climate scenarios. The framework can also be used to identify (i) regions under threat of productivity decline, and (ii) alternative crops and their locations that can cope better with changing climate. The methods and contents of the framework are presented by means of a case study developed in the South West Region of Victoria, Australia. The results can be used to assess land suitability in support of optimised crop allocations across a local region, and to underpin the development of a regional adaptation policy framework designed to reduce the vulnerability of the agriculture sector to the impacts of climate change.

Understanding response surface optimisation to the modeling of Astaxanthin extraction from a novel strain Thraustochytrium sp. S7

A fast growing, highly orange color pigmented strain of Thraustochytrids was isolated from New Zealand marine waters. This strain showed efficient utilization of glycerol as carbon source and produced significant amount of cell dry biomass (2.08gL-1), TFA (30.15% of dry cell weight), DHA (27.83% of TFA) and astaxanthin (131.56μgg-1). Astaxanthin is the dominant constituent in the carotenoid profile of Thraustochytrium sp. S7 and is an important antioxidant. Different cell disruption methods were applied for efficient astaxanthin extraction. Mechanical disruption of cells via ultrasonication resulted in the highest astaxanthin yield, from 26.78±1.25μgg-1 to 156.07±4.16μgg-1. Optimization of ultrasonication process using response surface methodology resulted into significant decrease in lysis time from 30min to 10min. This strain can be used for concurrent production of lipids and high value co-products such as DHA and astaxanthin.

Performance analysis using subsuming methods: an industrial case study

Large-scale object-oriented applications consist of tens of thousands of methods and exhibit highly complex runtime behaviour that is difficult to analyse for performance. Typical performance analysis approaches that aggregate performance measures in a method-centric manner result in thinly distributed costs and few easily identifiable optimisation opportunities. Subsuming methods analysis is a new approach that aggregates performance costs across repeated patterns of method calls that occur in the application's runtime behaviour. This allows automatic identification of patterns that are expensive and represent practical optimisation opportunities. To evaluate the practicality of this analysis with a real world large-scale object-oriented application we completed a case study with the developers of letterboxd.com - a social network website for movie goers. Using the results of the analysis we were able to rapidly implement changes resulting in a 54.8% reduction in CPU load and an 49.6% reduction in average response time.

Flexible transfer learning framework for bayesian optimisation

Bayesian optimisation is an efficient technique to optimise functions that are expensive to compute. In this paper, we propose a novel framework to transfer knowledge from a completed source optimisation task to a new target task in order to overcome the cold start problem. We model source data as noisy observations of the target function. The level of noise is computed from the data in a Bayesian setting. This enables flexible knowledge transfer across tasks with differing relatedness, addressing a limitation of the existing methods. We evaluate on the task of tuning hyperparameters of two machine learning algorithms. Treating a fraction of the whole training data as source and the whole as the target task, we show that our method finds the best hyperparameters in the least amount of time compared to both the state-of-art and no transfer method.

Heuristic approaches for multi-criteria optimisation in kidney exchange programs

The Kidney Exchange Problem (KEP) is an optimisation problem that was first discussed in Rapaport (1986) but has only more recently been the subject of much work by combinatorial optimisation re-searchers. This has been in parallel with its increased prevalence in the medical community. In the basic formulation of a KEP, each instance of the problem features a directed graph D = (V,A) . Each node i ∈ V represents an incompatible pair wherein the patient needs to trade kidneys with the patient of another incompatible pair. The goal is to find an optimal set of cycles such that as many patients as possible receive a transplant. The problem is further complicated by the imposition of a cycle-size constraint, usually considered to be 3 or 4. Kidney exchange programs around the world implement different algorithms to solve the allocation problem by matching up kidneys from potential donors to patients. In some systems all transplants are considered equally desirable, whereas in others, ranking criteria such as the age of the patient or distance they will need to travel are applied, hence the multi-criteria nature of the KEP. To address the multi-criteria aspect of the KEP, in this paper we propose a two-stage approach for the kidney exchange optimisation problem. In the first stage the goal is to find the optimal number of exchanges, and in the second stage the goal is to maximise the weighted sum of the kidney matches, subject to the added constraint that the number of exchanges must remain optimal. The idea can potentially be extended to multiple-objectives, by repeating the process in multiple runs. In our preliminary numerical experiments, we first find the maximum number of kidney matches by using an existing open source exact algorithm of Anderson et al. (2015). The solution will then be used as an initial solution for the stage two optimisation problem, wherein two heuristic methods, steepest ascent and random ascent, are implemented in obtaining good quality solutions to the objective of maximizing total weight of exchanges. The neighbourhood is obtained by two-swaps. It is our intention in the future to implement a varying neighbourhood scheme within the same two heuristic framework, or within other meta-heuristic framework.