Deep learning-based methods are nowadays widely applied for different tasks. However, application of these methods in modeling and prediction of the properties of chemical molecules has been less successful compared to other fields. This can be attributed both to the lack of data on chemical molecules, that could be used for model training, and to the imperfection of the employed molecular representations and deep learning methods. Aiming to tackle these problems, in the PhD project we propose to develop novel deep learning methods for the analysis of chemical molecules and prediction of their properties.
Applicants are expected to have either good knowledge and experience in chemistry/biochemistry, or skills in machine learning and neural networks.