Machine Learning for Materials Design and Discovery
The advent of data-centric approaches in the past decade has witnessed a paradigm shift in the way materials design and discovery has been pursued traditionally. This rapidly developing and vibrant scientific field, referred to as Materials Informatics, offers unparalleled opportunities by bringing together experts from diverse backgrounds, offers excitement and renewed hope in addressing difficult and grand challenge problems by introducing novel and transformative interdisciplinary research directions, as well as poses new challenges that call for clear standards to avoid hyperbole and promote scientifically meaningful pursuit of fundamental and applied materials research. This very-timely Special Topic aims to serve as a valuable forum for the community where scientists and practitioners in the field share their most recent findings, disruptive ideas and future outlook. The Special Topic will include original research articles, invited tutorials, and invited perspectives from experts in the field that highlight recent advances and set the scope for future research directions to address the most critical challenges in the broader field of materials science and engineering.
Topics covered include, but are not limited to:
- Efficient strategies for materials and molecular chemical space explorations
- Physics-informed versus physics-agnostic learning strategies
- Descriptor design, feature engineering and feature selection techniques
- Automated knowledge discovery and rule mining
- Natural language processing for developing materials databases
- Inverse design problems in materials science
- Multifidelity information fusion
- Automated learning of structure-property-processing mappings
- Learning from small and heterogeneous datasets
- Novel learning methods, validation strategies and uncertainty quantification
- Adaptive design, Bayesian optimization and active learning strategies in materials
- Surrogate modeling for materials multi-objective optimization problems
- Machine learning based strategies for force-fields development
- Reinforcement Learning for materials design and/or synthesis
- Best practices in materials informatics to promote and encourage reproducible results
- Success from failure: machine learning from negative results
Guest Editors
Rama Vasudevan – Oak Ridge National Laboratory
Ghanshyam Pilania – Los Alamos National Laboratory
Prasanna V. Balachandran – University of Virginia
Submission and acceptance criteria:
Manuscripts considered for publication in Journal of Applied Physics are expected to meet the journal’s standard of acceptance, i.e. to report on original and timely results that significantly advance understanding in the current status of contemporary applied physics; material that is exclusively review in nature is not considered for publication. Manuscripts submitted for consideration in this Special Topic must meet the same criteria and will undergo the journal’s standard peer-review process. The Editorial Team of the Journal of Applied Physics will issue final decisions on the submitted manuscripts.
For more information on the journal’s editorial policies, please click here.