The field of evolutionary algorithms (EAs) emerged in the area of computer science due to transfer of ideas from biology and developed independently for several decades, enriched with techniques from probability theor...
详细信息
During the past few decades, many evolutionary algorithms (EAs) together with the Constraint Handling Techniques (CHTs) have been developed to solve the constrained optimization problems (COPs). To obtain competitive ...
详细信息
The mission planning of agile earth observation satellite (AEOS) involves multiple objectives to be optimized simultaneously. Total profit, observed target number, averaged image quality, satellite usage balance and a...
详细信息
Nature features a plethora of extraordinary photonic architectures that have been optimized through natural evolution. While numerical optimization is increasingly and successfully used in photonics, it has yet to rep...
详细信息
Contemporary developments in the perception of our environment have caused a completely new pro- cess industry. Production processes now have to operate as optimal as possible with respect to different, often conflict...
详细信息
Contemporary developments in the perception of our environment have caused a completely new pro- cess industry. Production processes now have to operate as optimal as possible with respect to different, often conflicting, objectives as for instance optimality with respect to societal, environmental and eco- nomical aspects. In such case, no one optimal solution exists and decision makers need to resort to trade-off (or Pareto optimal) solutions. The problem of optimising a process with respect to different con- flicting objectives, is called a multi-objective optimisation problem (MOOP). MOOPs have gained a lot of attention in different applications throughout the past decade. MOOPs are however mathematically challenging problems and are generally solved via the use of dedicated algorithms. The two major al- gorithm categories are deterministic and stochastic algorithms. The former converts the MOOP in a set parameterised single-objective optimisation problems (SOOPs), while the latter tackles the MOOP in its entirety. The focus of this thesis is on the stochastic algorithms, and more specifically on the evolution- ary algorithms Non-dominated sorting genetic algorithm-II (NSGA-II) and Non-dominated sorting genetic algorithm-III (NSGA-III). These algorithms are widely acclaimed but still show two major shortcomings: (i) they are incapable of distinguishing between solutions based on their trade-off and distribution; (ii) they utilise a problem-irrelevant stopping criterion (i. e. reaching a pre-defined number of iterations). The lack of a trade-off based selection procedure results in a Pareto front that contains solutions that present no relevant information to the user. The flawed stopping criterion results in an unnecessary high com- putation time. To alleviate these shortcomings, the tM- and eventually tDOM-algorithms are developed. These algorithms can distinguish between solutions based on their trade-off via the implementation of t-domination. The tM-algorith
Distribution systems (DS) service restoration is a multi-objective, multi-constraint, combinatorial and non-linear optimization problem that must be quickly solved. Four multi-objective evolutionary algorithms (MOEAs)...
详细信息
Endovascular treatment for acute ischemic stroke, although an improvement upon previous treat- ments, does not always lead to a beneficial outcome. This thesis investigates the possibility of predicting treatment outc...
详细信息
Endovascular treatment for acute ischemic stroke, although an improvement upon previous treat- ments, does not always lead to a beneficial outcome. This thesis investigates the possibility of predicting treatment outcome with the use of evolutionary algorithms and decision trees. Three different evolutionary algorithms are designed to create decision trees which can predict outcome before treatment is administered. The first method is an evolutionary algorithm according to the general framework. Two additional extensions are proposed; one to generate oblique decision trees and one to generate fuzzy decision trees. The three methods are compared to the benchmark algorithm CART. The fuzzy and linear extensions both significantly outperform CART. However, due to insufficient accuracy (lower than 80%), the created decision trees can not be used as stand- alone method for outcome prediction. Nonetheless, the created decision trees can provide new insights in predictive value of the considered variables, and the created algorithm might be useful in bagging methods such as in a random forest.
Recent advancements in the machine learning algorithms, especially the development of Deep Neural Networks (DNNs) have transformed the landscape of Artificial Intelligence (AI). With every passing day, deep learning b...
详细信息
Recent advancements in the machine learning algorithms, especially the development of Deep Neural Networks (DNNs) have transformed the landscape of Artificial Intelligence (AI). With every passing day, deep learning based methods are applied to solve new problems with exceptional results. However true impact of AI could only be fully realized if it interacts with the real world and solves everyday problems. The everyday problem however, is new everyday and subject to increasingly changing requirements. The Deep Learning (DL) landscape today is incapable of solving these dynamic problems as the performance of DL today is heavily tied to the topology which is often task specific and hand-tuned by experts. Not only is rigidity of the solution the problem but also the high memory and compute requirements of DNNs to perform training on terabytes of data acts a huge barrier in bringing true intelligence to the edge which is the true portal to the 'real world'. NeuroEvolution (NE) are a class of algorithms that can circumvent this problem by 'learning on the fly'. These algorithms continuously interact with the environment and update their models based on how fruitful their last interaction proved. This way the solution is not tied to a topology and these algorithms do not need to perform memory and compute intensive backpropagation operations (BP) making them ideal for solving dynamic problems in a robust manner on the edge. However, the barrier to deployment of NE today is the lack of its widespread adoption and understanding of its compute behavior. This thesis attempts to lift that barrier by characterizing the compute and communication behavior a NE algorithm NEAT (NeuroEvolution of Augmenting Topologies) and is an attempt to propel further research in this direction. This Thesis also attempts to bring intelligence to the edge using a distributed system solution. This thesis demonstrates CLAN, Collaborative Learning using Asynchronous Neuro-evolution. It proposes tech
evolutionary optimization algorithms have been recently introduced as nonimaging optics design techniques. Unlike optimization of imaging systems, non sequential ray tracing simulations and complex non centred systems...
详细信息
ISBN:
(数字)9781510629349
ISBN:
(纸本)9781510629349
evolutionary optimization algorithms have been recently introduced as nonimaging optics design techniques. Unlike optimization of imaging systems, non sequential ray tracing simulations and complex non centred systems design must be considered, adding complexity to the problem. The Merit Function (MF) is a key element in the automatic optimization algorithm, nevertheless the selection of each objective's weight, {w(i)}, inside merit function needs a previous trial and error process for each optimization. The problem then is to determine appropriate weights value for each objective. In this paper we propose a new Dynamic Merit Function, DMF, with variable weight factors {w(i)(n)}. The proposed algorithm, automatically adapts weight factors, during the evolution of the optimization process. This dynamic merit function avoids the previous trial and error procedure selecting the right merit function and provides better results than conventional merit functions (CMF). Also we analyse the Multistart optimization algorithm applied in the flowline nonimaging design technique.
The usage of evolutionary algorithms for generating images has been researched for several decades now. The potential of this approach comes from the creative power of genetic operators and broad possibilities for aut...
详细信息
The usage of evolutionary algorithms for generating images has been researched for several decades now. The potential of this approach comes from the creative power of genetic operators and broad possibilities for automated evaluation of solutions. Individuals can be either evolved to resemble an existing image or other criteria such as artistic quality can be employed. Generating vector images to resemble raster models got a lot of attention in past years. It offers several benefits. Such images can be easily scaled without any loss of accuracy. Another advantage is the option to modify individual objects in an image separately. This aspect was, so far, being neglected. We want to reach full potential of evolved images by designing a suitable algorithm. Our method generates vector images similar to given raster model that are easily editable and have an interesting artistic overlap. We developed three techniques which differ in approach to individual representation, genetic operators, evaluation and overall style of results.
暂无评论