. "20"^^ . . "In this contribution is shown, what are possible perspectives of some selected methods of artificial intelligence in astrophysics, especially in prediction. Two methods and one special approach were selected here. The first method is so called neural networks. They are discussed in the first part of this participation including some simulations for demonstration. They are followed by discussion of fractal geometry and its possibility in time series processing. In the second part is shown how can be usedevolutionary algorithms for retrieval of suitable predictive models. Two new different algorithms were used for simulations described here. The first one was differential evolution (DE) and the second one was Self-Organizing Migrating Algorithm (SOMA). Both algorithms were used in the same way to find the best model whose response is comparable with given time series as much as possible. Problem was build up like problem of optimization where the cost function was based on difference between original t" . . "RIV/70883521:28110/01:00000087!RIV/2002/GA0/281102/N" . "HaPMK" . . "ARTIFICIAL INTELLIGENCE IN ASTROPHYSICS" . "[7671772FB4D3]" . . . "RIV/70883521:28110/01:00000087" . . . . "neural network, fractal geometry, prediction, solar activity, Elliott's wave, SOMA, DE, prediction, identification, evolution, migration, and self-organization"@en . "ARTIFICIAL INTELLIGENCE IN ASTROPHYSICS" . . . . . . . . . "2001-11-08+01:00"^^ . . "Zelinka, Ivan" . "ARTIFICIAL INTELLIGENCE IN ASTROPHYSICS"@en . . "28110" . "In this contribution is shown, what are possible perspectives of some selected methods of artificial intelligence in astrophysics, especially in prediction. Two methods and one special approach were selected here. The first method is so called neural networks. They are discussed in the first part of this participation including some simulations for demonstration. They are followed by discussion of fractal geometry and its possibility in time series processing. In the second part is shown how can be usedevolutionary algorithms for retrieval of suitable predictive models. Two new different algorithms were used for simulations described here. The first one was differential evolution (DE) and the second one was Self-Organizing Migrating Algorithm (SOMA). Both algorithms were used in the same way to find the best model whose response is comparable with given time series as much as possible. Problem was build up like problem of optimization where the cost function was based on difference between original t"@en . "Hv\u011Bzd\u00E1rna a planet\u00E1rium Mikul\u00E1\u0161e Kopern\u00EDka" . . . . . . . "v tisku" . "ARTIFICIAL INTELLIGENCE IN ASTROPHYSICS"@en . "Brno" . "673886" . "1"^^ . "Brno" . "P(GA102/00/0526), P(GA102/99/1292), Z(MSM 265200014)" . . "0"^^ . "1"^^ . . "0"^^ .