An image based prediction of the effective heat conductivity for highly heterogeneous microstructured materials is presented. The synthetic under consideration show different inclusion morphology, orientation, volume fraction and topology. property made exclusively on data with main emphasis being put 2-point spatial correlation function. This task implemented using both unsupervised supervised machine learning methods. First, a snapshot proper orthogonal decomposition (POD) used to analyze...

Abstract Two approaches are presented to improve the capabilities of machine learning models in multiscale modeling for microstructure homogenization (graphical abstract Fig. 1). The first approach features a Bayesian data mining scheme with human loop, halving prediction error compared [1] using four novel and efficient evaluate feature descriptors. second purely learning-driven utilizes convolutional neural networks, where we introduce module (the deep inception module) designed capture...

<title>Abstract</title> Two approaches are presented to improve the capabilities of machine learning models in multi- scale modeling for microstructure homogenization (graphical abstract Fig. 1). The first approach features a Bayesian data mining scheme with human loop, halving prediction error com- pared [1] using four novel efficient evaluate feature descriptors. second purely driven utilizes convolutional neural networks, where we introduce mod- ule, designed capture characteristics...

Beyond the generally deployed features for microstructure property prediction this study aims to improve machine learned by developing novel feature descriptors. Therefore, Bayesian infused data mining is conducted acquire samples containing characteristics inexplicable current set, and suitable descriptors describe these are proposed. The iterative development of resulted in 37 features, being able reduce error roughly one third. To further predictive model, convolutional neural networks...

Abstract In multiscale modeling, the response of macroscopic material is computed by considering behavior microscale at each point. To keep computational overhead low when simulating such high performance materials, an efficient, but also very accurate prediction microscopic utmost importance. Artificial neural networks are well known for their fast and efficient evaluation. We deploy fully convolutional networks, with one advantage being that, compared to directly predicting homogenized...

An image based prediction of the effective heat conductivity for highly heterogeneous microstructured materials is presented. The synthetic under consideration show different inclusion morphology, orientation, volume fraction and topology. property made exclusively on data with main emphasis being put 2-point spatial correlation function. This task implemented using both unsupervised supervised machine learning methods. First, a snapshot proper orthogonal decomposition (POD) used to analyze...