79 research outputs found
Parallelizing the Linkage Tree Genetic Algorithm and Searching for the Optimal Replacement for the Linkage Tree
No full textThe recently introduced Linkage Tree Genetic Algorithm (LTGA) has shown to exhibit excellent scalability on a variety of optimization problems. LTGA employs Linkage Trees (LTs) to identify and exploit linkage information between problem variables. In this work we present two parallel implementations of LTGA that enable us to leverage the computational power of a multi-processor architecture. These algorithm extensions for LTGA enable us to solve a problem that previously could not be solved, being the problem of finding high-quality predetermined linkage models that result in a better performance of LTGA for intricate problems by replacing the online-learned LTs. This is done by learning high-quality LTs offline by optimizing LTGAs performance as a function of static LTs. This results in a better performance of LTGA than with online-learned LTs as the problem complexity increases. A parameter-free implementation is used to search optimal subsets of linkage sets in the offline-learned LTs. This pruning of the LT results in a further performance improvement of the LTGA by, on average, removing about 50% of the linkage sets from the offline-learned LTs. This suggests that LTs contain redundancies that may possibly still be exploited to improve the performance of LTGA with online-learned LTs.AlgorithmicsSoftware TechnologyElectrical Engineering, Mathematics and Computer Scienc
Designing the Real-Valued Gene-pool Optimal Mixing Evolutionary Algorithm and Applying it to Substantially Improve the Efficiency of Multi-Objective Deformable Image Registration
No full textThe recently introduced Gene-pool Optimal Mixing Evolutionary Algorithm (GOMEA) for discrete variables has been shown to be able to efficiently and effectively exploit the decomposability of optimization problems, especially in a grey-box setting, in which a solution can be efficiently updated after a modification of a subset of its variables. GOMEA is considered to be state of the art, but currently no version of GOMEA for real-valued variables exists. In this thesis, we design a real-valued version of GOMEA, for both single-objective and multi-objective optimization. Our novel GOMEA variant is then applied to the Deformable Image Registration (DIR) problem, which was adapted to allow for efficient partial evaluations. DIR concerns the calculation of a deformation that transforms one image to another, and is of great importance for many medical applications. Experiments are performed to assess GOMEA’s performance in black-box and grey-box settings on a range of single-objective and multi-objective benchmark problems, including comparisons with it to the state-of-the-art real-valued optimization algorithm AMaLGaM. From the results of these experiments, we find that GOMEA performs substantially better on all considered single-objective and multi-objective benchmark problems in a grey-box setting, in terms of required time and number of evaluations. Moreover, the improvement becomes larger as problem dimensionality increases. In a black-box setting, GOMEA still performed better than AMaLGaM in terms of time, and comparable in terms of the number of evaluations. On DIR problems, GOMEA achieved solutions of similar quality while achieving a speed-up of up to a factor of 1600.Electrical Engineering, Mathematics and Computer ScienceSoftware TechnologyAlgorithmic
Improving the homogeneity of brachytherapy treatment plans generated by BRIGHT using a hotspot registration method based on connected component analysis
No full textIn this thesis BRIGHT, a multi-objective evolutionary algorithm for the creation of treatment plans for high-dose rate brachytherapy for prostate cancer, is extended with a new objective to mitigate the formation of high dose contiguous volumes, i.e. hotspots. Multiple new objectives are tested on their performance to reduce hotspots, while minimally impacting existing objectives
Multi-objective optimization of wind farm layouts – Complexity, constraint handling and scalability
No full textAbstractCurrently, Offshore Wind Farms (OWFs) are designed to achieve high turbine density so as to reduce costs. However, due to wake interferences, densely packing turbines reduces energy production. Having insight into optimized trade-offs between energy production, capital investment and operational costs would be valuable to OWFs designers. To obtain this insight, the design of OWFs should be formulated as a multi-objective optimization problem. How to best solve a Multi-Objective Wind Farm Layout Optimization Problem (MOWFLOP) is however still largely an open question. It is however known that evolutionary algorithms (EAs) are among the state-of-the-art for solving multi-objective optimization problems. This work studies the different features that an MO Evolutionary Algorithm (MOEA) should have and which Constraint-Handling Techniques (CHTs) are suitable for solving MOWFLOP. We also investigate the relation between problem dimensionality/complexity and the degrees of freedom offered by different turbine-placement grid resolutions. Finally, the influence of problem size on algorithm performance is studied. The performance of two variants of the recently introduced Multi-Objective Gene-pool Optimal Mixing Evolutionary Algorithm (MOGOMEA) is compared with a traditional and a novel version of the Nondominated Sorting Genetic Algorithm II (NSGA-II). Five CHTs were used to assess which technique provides the best results. Results on a case study with different OWF areas demonstrate that one variant of MOGOMEA outperforms the NSGA-II for all tested problem sizes and CHTs
A Multi-Objective Optimization Framework for the Design of Offshore Wind Farms
No full textDC systems, Energy conversion & Storag
Neutrino Mass Matrix Textures: A Data-driven Approach
No full textWe analyze the neutrino mass matrix entries and their correlations in a probabilistic fashion, constructing probability distribution functions using the latest results from neutrino oscillation fits. Two cases are considered: the standard three neutrino scenario as well as the inclusion of a new sterile neutrino that potentially explains the reactor and gallium anomalies. We discuss the current limits and future perspectives on the mass matrix elements that can be useful for model building.We analyze the neutrino mass matrix entries and their correlations in a probabilistic fashion, constructing probability distribution functions using the latest results from neutrino oscillation fits. Two cases are considered: the standard three neutrino scenario as well as the inclusion of a new sterile neutrino that potentially explains the reactor and gallium anomalies. We discuss the current limits and future perspectives on the mass matrix elements that can be useful for model building.We analyze the neutrino mass matrix entries and their correlations in a probabilistic fashion, constructing probability distribution functions using the latest results from neutrino oscillation fits. Two cases are considered: the standard three neutrino scenario as well as the inclusion of a new sterile neutrino that potentially explains the reactor and gallium anomalies. We discuss the current limits and future perspectives on the mass matrix elements that can be useful for model building
Hybridizing Hypervolume based Evolutionary Algorithms and Gradient Descent by Dynamic Resource Allocation
No full textGPU-Accelerated GOMEA: Solving the max-cut problem by large-scale parallelisation of GOMEA using GPGPU
No full textWith the advances in General-Purpose computing on Graphics Processing Units (GPGPU), it is worthwhile to explore whether other areas in the field of Artificial Intelligence (AI) can reap the benefits. One such area is Evolutionary Algorithms (EAs), which—among other processes—involves the repetitive exchange of genes among individuals. This repetitive nature aligns with our intuition for parallel optimisation, precisely what GPGPU is designed for. Currently, the state-of-the-art approach in EA is known as Gene-pool Optimal Mixing Evolutionary Algorithm (GOMEA), which capitalises on the information embedded within the population by computing the linkage between genes across the entire population. However, when it comes to parallelising the exchange of complete linkage sets, particularly in the context of our specific problem of interest, the challenge becomes more intricate.In the case of our problem, known as Max-cut, there are dependencies between genes that must be considered when constructing parallel sets of linkage sets, referred to as packages. We propose three solutions: contamination, revision, and association. Contamination fully utilises parallel capabilities but deviates from the concept of linkage sets. Revision constructs the linkage sets as described by GOMEA, but keeps the dependencies between linkage sets within a package untouched. Association on the other hand attempts to resolve the dependencies by generating a dependency graph to create the set of packages.From our experiments, we can conclude that parallel acceleration using GPGPU is roughly on par with—and sometimes even outperforms—its non-parallelised counterpart. Out of the three solutions, it is evident that association demonstrates the most promising performance profile in terms of approaching the optimal solution. However, the performance falls significantly short of matching the capabilities exhibited by GOMEA. Furthermore, all of the solutions face a significant burden when evaluating the fitness for each exchanged linkage set. An option to consider as an extension to the current setup is known as partial evaluation, although the performance exhibited by contamination implies that simplicity could be the key to success. Further exploration of the acceleration process using widely employed parallel operators—such as those found in linear algebra—has the potential to yield valuable insights for enhancing performance.Applied Mathematic
Optimal Mixing Evolutionary Algorithms for Large-Scale Real-Valued Optimization: Including Real-World Medical Applications
No full textIn recent years, the use of Artificial Intelligence (AI) has become prevalent in a large number of societally relevant, real-world problems, e.g., in the domains of engineering and health care. The field of Evolutionary Computation (EC) can be considered to be a sub-field of AI, concerning optimization using Evolutionary Algorithms (EAs), which are population-based (meta-)heuristics that employ the Darwinian principles of evolution, i.e., variation and selection. Such EAs are historically mainly considered for the optimization of difficult, non-linear problems in a Black-Box Optimization (BBO) setting, because EAs can effectively optimize such problems even when very little is known about the optimization problem and its structure. This is in contrast to optimization methods that are specifically designed for certain problems of which the definition and structure are known, i.e., a White-Box Optimization (WBO) setting
Design and Application of Gene-pool Optimal Mixing Evolutionary Algorithms for Genetic Programming
No full textMachine learning is impacting modern society at large, thanks to its increasing potential to effciently and effectively model complex and heterogeneous phenomena. While machine learning models can achieve very accurate predictions in many applications, they are not infallible. In some cases, machine learning models can deliver unreasonable outcomes. For example, deep neural networks for self-driving cars have been found to provide wrong steering directions based on the lighting conditions of street lanes (e.g., due to cloudy weather). In other cases, models can capture and reflect unwanted biases thatwere concealed in the training data. For example, deep neural networks used to predict likely jobs and social status of people based on their pictures, were found to consistently discriminate based on gender and ethnicity–this was later attributed to human bias in the labels of the training data.Algorithmic
- …
