Time: Mar’20 - May’20
Course: 06-801 Data Science and Machine Learning in ChemE
Source Code: Github Repository
- Automatically generated atomic structures were transformed into machine-learnable input features (SOAP descriptors) which were then used to predict the binding energies of Carbon.
- Developed a program capable of automatically generating atomistic structures using user-specified elements reducing required manual entry time by 90%
- Constructed and trained a model capable of predicting binding energies with an average error of less than 10%.