Moleadapt: A recommender system for molecular property adaptation
| Title: |
Moleadapt: A recommender system for molecular property adaptation |
| DNr: |
NAISS 2023/22-161 |
| Project Type: |
NAISS Small Compute |
| Principal Investigator: |
Michael Welle <mwelle@kth.se> |
| Affiliation: |
Kungliga Tekniska högskolan |
| Duration: |
2023-02-08 – 2024-03-01 |
| Classification: |
10405 |
| Keywords: |
|
Abstract
Moleadapt is a machine learning-based recommender system that enables the user to specify a certain property of molecules that should be adapted to user specifications. The system takes in the wanted parameter and outputs molecular models that best fulfill the given criteria while being as similar to the starting molecule as possible.
Using a dataset of thousands of molecule specifications with given parameters our model designed a representation space that helps to recommend promising candidates to the user.
