.png){kind=icon}
.png){kind=icon}
Return
MATCALO: Knowledge-enabled machine learning in materials science
.png){kind=icon}
OA
AI
DOI：10.1016/j.commatsci.2019.03.005.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Industrial and technological innovations constantly call for the development of materials that meet specific and novel requirement profiles. However, the engineering of materials with the desired properties is a time-, cost-, and labor-intensive process as changing the composition and the parameterization of processing steps yields an unimaginably large search space. It is therefore crucial for the future engineering of materials to be conducted in a more goal-directed and more efficient way to generate informative hypotheses revealing particularly promising experiments and thus guiding the exploration process in a more informed fashion. In this paper, we present MATCALO, an intelligent, cognitive assistant system that supports materials scientists in developing novel materials. MATCALO combines modern machine learning techniques with machine-interpretable semantic knowledge in order to model representations of relationships between materials, processes and properties and allow reasoning about them. We showcase a prototypical approach for generating such hypotheses by reverse-querying regression models learned from experimental data. However, a workable set of experiments alone generally does not suffice as it will never cover an adequately large area of the search space, which led us to the conclusion that additional background knowledge is required to build more reliable models. The readers can assure themselves of the feasibility of this approach by testing the web-based interface using the MATCALO system which will be made available to the broad community. Given a requirement profile, the system generates hypotheses on how these criteria can be achieved.
Keywords:
Computational materials science
Artificial intelligence
Materials informatics
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
3.3
Papers:
1.3W
Citations:
3.6W