This position aims at fundamental developments of machine learning methods to design new materials under uncertain conditions. Materials Science sits between fundamental- and applied sciences because a new material needs to consider the entire process-structure-property chain.
- Unfortunately, real materials have uncertain properties and under the same conditions, two macroscopically identical materials exhibit stochastic values for the same properties. Machine Learning may offer methods to predict material characteristics under these uncertain conditions.
- The goal of this project is to develop Machine Learning models that incorporate physical knowledge and constraints to not only predict these material characteristics, but also their uncertainty. You will work on the necessary concepts and methodologies to learn models that can (1) predict material properties and (2) quantify the certainty in these properties.
- These models should adhere to physical constraints and have to be fitted to poorly sampled data as well as very large datasets. A possible approach could be to combine the concepts of deep learning with the Bayesian formalism, so that the very flexible deep learning models can be equipped with solid probabilistic reasoning.
- The recently established “MACHINA” Artificial Intelligence Laboratory at TU Delft comprise members of the Faculty of Mechanical, Maritime and Materials Engineeringand of the Faculty of Electrical Engineering, Mathematics and Computer Science.
- A university Master of Science degree in computer science, electrical engineering, mathematics, physics or related area.
- The ability to translate the requirements imposed by the applications of the group into a Machine Learning formulation, and translate back the Machine Learning results and limitations into the application relevant quantities.
- Good writing and presentation skills in English, and certainly proficient in programming.
- Prior knowledge in mechanical engineering or materials science is not required, but there should be a strong interest in cooperating with colleagues in these areas.
How to Apply
To apply, please email a letter of motivation, a detailed CV, an abstract of the Master thesis (1 page), preferred starting date, names of referees and MSc transcripts/Diploma (in English) in a single pdf file preferably before September 1st, 2020.
- For information about this vacancy, you can contact Dr. David Tax, Assistant Professor (Intelligent Systems), email: M.J.Tax@tudelft.nl,
- Miguel Bessa, Assistant Professor (Materials Science and Engineering), email: M.A.Bessa@tudelft.nl,
|Organization||Delft University of Technology( TU Delft)|
|Type of position||Doctoral position|
|Subject areas||Master of Science, mathematics, physics computer science, electrical engineering|
|Deadline||1 September , 2020.|