Virginia Commonwealth University
Information Architecture for a Chemical Modeling Knowledge Graph
Abstract
dc:description.abstractMachine learning models for chemical property predictions are high dimension design challenges spanning multiple disciplines. Free and open-source software libraries have streamlined the model implementation process, but the design complexity remains. In order better navigate and understand the machine learning design space, model information needs to be organized and contextualized. In this work, instances of chemical property models and their associated parameters were stored in a Neo4j property graph database. Machine learning model instances were created with permutations of dataset, learning algorithm, molecular featurization, data scaling, data splitting, hyperparameters, and hyperparameter optimization techniques. The resulting graph contains over 83,000 nodes and 4 million edges and can be explored with interactive visualization software. The structure of the property graph is centered around models and molecules which enables efficient and intuitive inter- and intra-model evaluation. We use a curated lipophilicity dataset to demonstrate graph use cases. Difficult to predict molecules were identified across multiple models simultaneously. Powerful and expressive graph queries were implemented to identify molecular fragments that were both prevalent and associated with high lipophilicity prediction error.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- Thesis
- Discipline thesis:degree_discipline
- Chemical and Life Science Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Luxon, Adam R
- Contributors dc:contributor
-
- Dr. James Ferri
- Dr. David Tyler McQuade
- Mr. William Glandorf
Subjects
dc:subject × 10Rights
dc:rights- Statement dc:rights
-
- © The Author
Identifiers
dc:identifier.*- Identifier
- https://scholarscompass.vcu.edu/etd/6585