The development of new materials, incorporation of new functionalities, and even the description of well-studied materials strongly depends on the capability of individuals to deduce complex structure-property relationships. A significant challenge in this field remains the “curse of dimensionality.” Even for the characterization of materials with a rather moderate number of constituents and relevant scales, often high-dimensional parameter spaces are required. And although in theory an appropriate set of parameters, describing, e.g., materials composition and processing conditions, uniquely determines (at least in a statistical sense) the properties of a material, the identification of the relationship between parameters and properties is non-trivial. In general, this requires pointwise sampling. Relying only on experiments to this end is typically prohibitively expensive, given the often high-dimensional parameter space of interest. Thus, a combination of experimental and computational approaches has received increasing attention. The complex interdependencies in the resulting data sets can be studied using machine-learning approaches: artificial neural networks and data-driven approaches can significantly help to identify, approximate, and visualize parameter-property relationships of interest. This way, they can replace, at least in part, time-consuming modeling, simulations, or experiments, and can serve to accelerate data generation or scale bridging.
This Research Topic intends to publish contributions on current ideas and novel concepts for the advancement of machine learning, data mining, and data driven-approaches in the context of the design of materials and materials processing. Both papers on general methods, as well as their applications to decoding the complex relationships along the chain of composition-processing-structure-mechanical properties, are highly welcome. The replacement of experiments with validated and quantitative predictive computational methods, the identification of dependencies and mechanisms of deformation, damage, degradation, and fracture from big data, as well as the representation of high-dimensional relationships by computer models, are of particular relevance.
Topics of interest are:
• Automatized data generation from experiments or computational methods and their use for decoding hidden information
• Design of the distribution of data points aimed at maximizing information content
• Data mining, machine learning, artificial neural networks, data driven computing
• Treatment of scatter, outliers, and uncertainties
• Applications, covering for example constitutive modelling, property prediction, scale bridging, optimization, inverse problems
The development of new materials, incorporation of new functionalities, and even the description of well-studied materials strongly depends on the capability of individuals to deduce complex structure-property relationships. A significant challenge in this field remains the “curse of dimensionality.” Even for the characterization of materials with a rather moderate number of constituents and relevant scales, often high-dimensional parameter spaces are required. And although in theory an appropriate set of parameters, describing, e.g., materials composition and processing conditions, uniquely determines (at least in a statistical sense) the properties of a material, the identification of the relationship between parameters and properties is non-trivial. In general, this requires pointwise sampling. Relying only on experiments to this end is typically prohibitively expensive, given the often high-dimensional parameter space of interest. Thus, a combination of experimental and computational approaches has received increasing attention. The complex interdependencies in the resulting data sets can be studied using machine-learning approaches: artificial neural networks and data-driven approaches can significantly help to identify, approximate, and visualize parameter-property relationships of interest. This way, they can replace, at least in part, time-consuming modeling, simulations, or experiments, and can serve to accelerate data generation or scale bridging.
This Research Topic intends to publish contributions on current ideas and novel concepts for the advancement of machine learning, data mining, and data driven-approaches in the context of the design of materials and materials processing. Both papers on general methods, as well as their applications to decoding the complex relationships along the chain of composition-processing-structure-mechanical properties, are highly welcome. The replacement of experiments with validated and quantitative predictive computational methods, the identification of dependencies and mechanisms of deformation, damage, degradation, and fracture from big data, as well as the representation of high-dimensional relationships by computer models, are of particular relevance.
Topics of interest are:
• Automatized data generation from experiments or computational methods and their use for decoding hidden information
• Design of the distribution of data points aimed at maximizing information content
• Data mining, machine learning, artificial neural networks, data driven computing
• Treatment of scatter, outliers, and uncertainties
• Applications, covering for example constitutive modelling, property prediction, scale bridging, optimization, inverse problems