Neural networks for computationally expensive problems
Optimointiin tarvitaan usein simulaattoria, jotka voivat olla laskennallisesti raskaita. Simulaattorit voidaan korvata sijaismalleilla, jotka ovat nopeampi laskea ja voivat olla lähes yhtä tarkkoja kuin simulaattorit. Tässä työssä tarkastelemme tarkemmin yhtä sijaismalli, neuroverkkoja. Valmistelemme sijaismalli avusteista optimointia rakentamalla, opettamalla ja validoimalla erilaisia neuroverkkoja sijaismalliksi. Lisäksi vertailemme eri data samplaustekniikoilla generoitujen opetusdatojen vaikutusta neuroverkkojen approksimointitarkkuuteen. Optimization often involves usage of a simulator, which can be computationally expensive to use. Simulators can be replaced by surrogate models, which are computationally cheaper and can be almost as accurate as the simulators. In this thesis we consider closer a surrogate model, namely neural networks. We prepare surrogate assisted optimization by building, training and validating different neural network models for a surrogate model. In addition we compare how different training data sets, which are generated by different data sampling techniques, effect the generalization accuracy of neural networks.
Asiasanat
Metadata
Näytä kaikki kuvailutiedotKokoelmat
- Pro gradu -tutkielmat [28121]
Samankaltainen aineisto
Näytetään aineistoja, joilla on samankaltainen nimeke tai asiasanat.
-
Handling expensive multiobjective optimization problems with evolutionary algorithms
Chugh, Tinkle (University of Jyväskylä, 2017)Multiobjective optimization problems (MOPs) with a large number of conflicting objectives are often encountered in industry. Moreover, these problem typically involve expensive evaluations (e.g. time consuming simulations ... -
Approximation method for computationally expensive nonconvex multiobjective optimization problems
Haanpää, Tomi (University of Jyväskylä, 2012) -
On approaches for solving computationally expensive multiobjective optimization problems
Tabatabaei, Seyed Mohammad Mehdi (University of Jyväskylä, 2016)In this thesis, we consider solving computationally expensive multiobjective optimization problems that take into account the preferences of a decision maker (DM). The aim is to support the DM in identifying the most ... -
A survey on handling computationally expensive multiobjective optimization problems with evolutionary algorithms
Chugh, Tinkle; Sindhya, Karthik; Hakanen, Jussi; Miettinen, Kaisa (Springer, 2019)Evolutionary algorithms are widely used for solving multiobjective optimization problems but are often criticized because of a large number of function evaluations needed. Approximations, especially function approximations, ... -
On solving computationally expensive multiobjective optimization problems with interactive methods
Ojalehto, Vesa (University of Jyväskylä, 2014)
Ellei toisin mainittu, julkisesti saatavilla olevia JYX-metatietoja (poislukien tiivistelmät) saa vapaasti uudelleenkäyttää CC0-lisenssillä.