Share This Author
Deep learning for molecular design—a review of the state of the art
Several important high level themes emerge, including the shift away from the SMILES string representation of molecules towards more sophisticated representations such as graph grammars and 3D representations, the importance of reward function design, the need for better standards for benchmarking and testing, and the benefits of adversarial training and reinforcement learning over maximum likelihood based training.
Deep learning for molecular generation and optimization - a review of the state of the art
- D. Elton, Zois Boukouvalas, M. Fuge, Peter W. Chung
- Computer ScienceMolecular Systems Design & Engineering
- 11 March 2019
A recent groundswell of work which uses deep learning techniques to generate and optimize molecules and how these techniques improve the quality of existing molecules is reviewed.
Applying machine learning techniques to predict the properties of energetic materials
- D. Elton, Zois Boukouvalas, Mark S Butrico, M. Fuge, Peter W. Chung
- Computer Science, ChemistryScientific Reports
- 15 January 2018
This work presents a comprehensive comparison of machine learning models and several molecular featurization methods - sum over bonds, custom descriptors, Coulomb matrices, Bag of Bonds, and fingerprints - and concludes that the best featurizing was sum over bond counting, and the best model was kernel ridge regression.
Machine Learning of Energetic Material Properties
It is determined that even when using a small training set, non-linear regression methods may create models within a useful error tolerance for screening of materials.
Locally Optimizable Joint Embedding Framework to Design Nitrogen‐rich Molecules that are Similar but Improved
- Sangeeth Balakrishnan, Francis G. VanGessel, Zois Boukouvalas, Brian C. Barnes, M. Fuge, Peter W. Chung
- BiologyMolecular informatics
- 28 April 2021
A deep learning method that combines a generative model with a property prediction model to fuse small data of one class of molecules with larger data in another class, demonstrating that fusing and joint embedding with plentiful low nitrogen molecular data can produce higher generative performance than using the scarce data alone.
Independent Vector Analysis for Molecular Data Fusion: Application to Property Prediction and Knowledge Discovery of Energetic Materials
- Zois Boukouvalas, Monica Puerto, D. Elton, Peter W. Chung, M. Fuge
- Computer Science28th European Signal Processing Conference…
- 24 January 2021
This work proposes a data fusion framework that uses Independent Vector Analysis to uncover underlying complementary information contained in different molecular featurization methods and generates a low dimensional set of features—molecular signatures—that can be used for the prediction of molecular properties and for knowledge discovery.
Using natural language processing techniques to extract information on the properties and functionalities of energetic materials from large text corpora
This work explores how techniques from natural language processing and machine learning can be used to automatically extract chemical insights from large collections of documents and compares the utility of two popular word embeddings.
Independent Vector Analysis for Data Fusion Prior to Molecular Property Prediction with Machine Learning
This work proposes a data fusion framework that uses Independent Vector Analysis to exploit underlying complementary information contained in different molecular featurization methods, bringing us a step closer to automated feature generation.
Chemically driven energetic molecular ferroelectrics
This work designs energetic molecular ferroelectrics consisting of imidazolium cations (energetic ion) and perchlorate anions (oxidizer) and describes its thermal wave energy conversion with a specific power of 1.8 kW kg−1, suggesting a polarization-dependent heat transfer and specific power suggests the role of electron-phonon interaction in tuning energy density of energetic Molecular ferroElectrics.
Assessing the trade-off between prediction accuracy and interpretability for topic modeling on energetic materials corpora
This work studies the trade-oﬀ between prediction accuracy and interpretability by implementing three document embedding methods that vary in computational complexity that provide local interpretability model-agnostic explanations of each prediction.