489 research outputs found
Predicting wildfire spreading through a hexagonal cellular automata model
No full textAs it is well known forest fires can present serious risk to people and can have enormous environmental impact. Therefore researchers and land managers are increasingly interested in effective tools for use in scientific analyses, management and fighting operations. On the other hand forest fires are complex phenomena that need an interdisciplinary approach. In this paper the paradigm of Cellular Automata was applied and a model was projected to simulate the evolution of forest fires. The adopted method involves the definition of local rules, mainly based on fire spread relationships originally developed by Rothermel in 1972, from which the global behaviour of the system can emerge. The preliminary results show that the model could be applied for forest fire prevention, the production of risk scenarios and the evaluation of the forest fire environmental impact
An Effective Approach for Adapting the Size of Subcomponents in Large-Scale Optimization with Cooperative Coevolution
No full textThe performance of cooperative co-evolutionary algorithms for large-scale global optimization (LSGO) can be significantly affected by the adopted problem decomposition. This study investigates a new adaptive Cooperative Coevolutionary algorithm in which several decompositions are concurrently applied during short learning phases. Moreover, the study includes some experimental results on a set of LSGO problems and a comparison with a recent approach based on reinforcement-learning. According to the numerical results, the proposed adaptive approach can provide a superior search efficiency on several benchmark functions
A Cooperative Coevolutionary Differential Evolution Algorithm with Adaptive Subcomponents
No full textAbstractThe performance of cooperative co-evolutionary (CC) algorithms for large-scale continuous optimization is significantly affected by the adopted decomposition of the search space. According to the literature, a typical decomposition in case of separable problems consists of adopting equally sized subcomponents for the whole optimization process (i.e. static decomposition). Such an approach is also often used for non-separable problems, together with a random-grouping strategy. More advanced methods try to determine the optimal size of subcomponents during the optimization process using reinforcement-learning techniques. However, the latter approaches are not always suitable in this case because of the non-stationary and history-dependent nature of the learning environment. This paper investigates a new CC algorithm, based on Differential Evolution, in which several decompositions are applied in parallel during short learning phases. The experimental results on a set of large-scale optimization problems show that the proposed method can lead to a reliable estimate of the suitability of each subcomponent size. Moreover, in some cases it outperforms the best static decomposition
Towards a visual graph-based story outline authoring
No full textIn this paper we present an authoring tool for collaborative visual creation of story outlines. A visual graph-based approach was adopted in order to allow story outlining of a wide range of types and genres of stories, while offering a story representation paradigm suitable for multi-author collaboration and for the integration and organisation of external sources of documentation of the story. Such a paradigm of story representation is also effective for an easy and semiautomatic rapid creation of story prototypes which can be obtained by mean of a specifically developed templating engine
Metaheuristics for Continuous Optimization of High-Dimensional Problems: State of the Art and Perspectives
No full textExploiting Spatio-temporal Data for the Multiobjective Optimization of Cellular Automata Models
No full textEnhancing the Firefly Algorithm through a Cooperative Coevolutionary Approach: an Empirical Study on Benchmark Optimization Problems
No full text
Enhancing Cooperative Coevolution with Surrogate-Assisted Local Search
No full textIn recent years, an increasing effort has been devoted to the study of metaheuristics suitable for large-scale global optimization in the continuous domain. However, so far the optimization of high-dimensional functions that are also computationally expensive has attracted little research. To address such an issue, this chapter describes an approach in which fitness surrogates are exploited to enhance local search (LS) within the low-dimensional subcomponents of a cooperative coevolutionary (CC) optimizer. The chapter also includes a detailed discussion of the related literature and presents a preliminary experimentation based on typical benchmark functions. According to the results, the surrogate-assisted LS within subcomponents can significantly enhance the optimization ability of a CC algorithm
- …
