Machine learning modelling of chemical reaction characteristics: yesterday, today, tomorrow

Abstract Chemical compound space refers to the vast set of all possible chemical compounds, estimated to contain 1060 molecules. While intractable as a whole, modern machine learning (ML) is increasingly capable of accurately predicting molecular properties in important subsets. Here, we therefore engage in the ML-driven study of even larger reaction space. Central to chemistry as a science of transformations, this space contains all possible chemical reactions. As an important basis for ‘reactive’ ML, we establish a first-principles database (Rad-6) containing closed and open-shell organic molecules, along with an associated database of chemical reaction energies (Rad-6-RE). We show that the special topology of reaction spaces, with central hub molecules involved in multiple reactions, requires a modification of existing compound space ML-concepts. Showcased by the application to methane combustion, we demonstrate that the learned reaction energies offer a non-empirical route to rationally extract reduced reaction networks for detailed microkinetic analyses.

Download Full-text

Molecular dynamics simulations of shock loading of nearly fully dense granular Ni–Al composites

Physical Chemistry Chemical Physics ◽

10.1039/c9cp02920f ◽

2019 ◽

Vol 21 (36) ◽

pp. 20252-20261

Author(s):

Yongnan Xiong ◽

Xiaofan Li ◽

Shifang Xiao ◽

Huiqiu Deng ◽

Bowen Huang ◽

...

Keyword(s):

Molecular Dynamics ◽

Molecular Dynamics Simulations ◽

Chemical Reaction ◽

Shock Loading ◽

Shock Propagation ◽

Inhomogeneous Deformation ◽

Reaction Characteristics ◽

Dynamics Simulations

We used molecular dynamics simulations to study the shock propagation, inhomogeneous deformation, and initiation of the chemical reaction characteristics of nearly fully dense reactive Ni–Al composites.

Download Full-text

Summit: Benchmarking Machine Learning Methods for Reaction Optimisation

10.26434/chemrxiv.12939806.v1 ◽

2020 ◽

Author(s):

Kobi Felton ◽

Jan Rittig ◽

Alexei Lapkin

Keyword(s):

Machine Learning ◽

Experimental Data ◽

Open Source ◽

Chemical Reaction ◽

Learning Strategy ◽

New Products ◽

Fine Chemicals ◽

Optimal Solutions ◽

Machine Learning Methods ◽

Open Source Framework

In the fine chemicals industry, reaction screening and optimisation are essential to development of new products. However, this screening can be extremely time and labor intensive, especially when intuition is used. Machine learning offers a solution through iterative suggestions of new experiments based on past experimental data, but knowing which machine learning strategy to apply in a particular case is still difficult. Here, we develop chemically-motivated virtual benchmarks for reaction optimisation and compare several strategies on these benchmarks. The benchmarks and strategies are encompassed in an open source framework named Summit. The results of our tests show that Bayesian optimisation strategies perform very well across the types of problems faced in chemical reaction optimisation, while many strategies commonly used in reaction optimisation fail to find optimal solutions.

Download Full-text

Chemical Reaction Spectrum

10.26434/chemrxiv.7240628.v1 ◽

2018 ◽

Author(s):

Jianchao Lee ◽

Jianghong Li ◽

Qiannan Duan ◽

Sifan Bi ◽

Ruen Luo ◽

...

Keyword(s):

Artificial Intelligence ◽

Data Acquisition ◽

Chemical Reaction ◽

Chemical Characterization ◽

Important Data ◽

Data Set ◽

Characterization Methods ◽

Graphical Data ◽

Reaction Characteristics ◽

Reaction Spectrum

We proposed a new method of chemical reaction spectrum (CRS) in terms of chemical characterization, and established a method to fulfill it by combining with 3D chemical printing technology and 2D sampling. The CRS can provide a graphical data set for pure or mixed substances, which can comprehensively describe the reaction characteristics of the research object. Compared with common characterization methods (NMR, UV/vis, IR, Raman, GC or LC), it is more capable of revealing chemical behaviors enough, and is much lower in cost. It is expected to be an important data acquisition approach for the application of artificial intelligence in the field of chemistry in the future.

Download Full-text

Improving machine learning performance on small chemical reaction data with unsupervised contrastive pretraining

10.26434/chemrxiv-2021-xr8tf-v2 ◽

2022 ◽

Author(s):

Mingjian Wen ◽

Samuel M. Blau ◽

Xiaowei Xie ◽

Shyam Dwaraknath ◽

Kristin A. Persson

Keyword(s):

Machine Learning ◽

Chemical Reaction ◽

Chemical Space ◽

Relevant Information ◽

Learning Performance ◽

Reaction Data ◽

Unlabelled Data ◽

Graph Neural Networks ◽

Small Chemical ◽

Fine Tune

Machine learning (ML) methods have great potential to transform chemical discovery by accelerating the exploration of chemical space and drawing scientific insights from data. However, modern chemical reaction ML models, such as those based on graph neural networks (GNNs), must be trained on a large amount of labelled data in order to avoid overfitting the data and thus possessing low accuracy and transferability. In this work, we propose a strategy to leverage unlabelled data to learn accurate ML models for small labelled chemical reaction data. We focus on an old and prominent problem—classifying reactions into distinct families—and build a GNN model for this task. We first pretrain the model on unlabelled reaction data using unsupervised contrastive learning and then fine-tune it on a small number of labelled reactions. The contrastive pretraining learns by making the representations of two augmented versions of a reaction similar to each other but distinct from other reactions. We propose chemically consistent reaction augmentation methods that protect the reaction center and find they are the key for the model to extract relevant information from unlabelled data to aid the reaction classification task. The transfer learned model outperforms a supervised model trained from scratch by a large margin. Further, it consistently performs better than models based on traditional rule-driven reaction fingerprints, which have long been the default choice for small datasets. In addition to reaction classification, the effectiveness of the strategy is tested on regression datasets; the learned GNN-based reaction fingerprints can also be used to navigate the chemical reaction space, which we demonstrate by querying for similar reactions. The strategy can be readily applied to other predictive reaction problems to uncover the power of unlabelled data for learning better models with a limited supply of labels.

Download Full-text

Binary chemical reaction optimization based feature selection techniques for machine learning classification problems

Expert Systems with Applications ◽

10.1016/j.eswa.2020.114169 ◽

2020 ◽

pp. 114169

Author(s):

P.C. Srinivasa Rao ◽

A.J. Sravan Kumar ◽

Quamar Niyaz ◽

Paheding Sidike ◽

Vijay K Devabhaktuni

Keyword(s):

Machine Learning ◽

Feature Selection ◽

Chemical Reaction ◽

Classification Problems ◽

Chemical Reaction Optimization ◽

Machine Learning Classification ◽

Reaction Optimization ◽

Feature Selection Techniques

Download Full-text

Predicting Chemical Reaction Barriers with a Machine Learning Model

Catalysis Letters ◽

10.1007/s10562-019-02705-x ◽

2019 ◽

Vol 149 (9) ◽

pp. 2347-2354 ◽

Cited By ~ 17

Author(s):

Aayush R. Singh ◽

Brian A. Rohr ◽

Joseph A. Gauthier ◽

Jens K. Nørskov

Keyword(s):

Machine Learning ◽

Chemical Reaction ◽

Learning Model ◽

Machine Learning Model ◽

Reaction Barriers

Download Full-text

Deep Reinforcement Learning of Transition States

Physical Chemistry Chemical Physics ◽

10.1039/d0cp06184k ◽

2021 ◽

Author(s):

Jun Zhang ◽

Yao-Kun Lei ◽

Zhen Zhang ◽

Xu Han ◽

Maodong Li ◽

...

Keyword(s):

Machine Learning ◽

Molecular Dynamics ◽

Reinforcement Learning ◽

Transition State ◽

Chemical Reaction ◽

Reaction Mechanisms ◽

Md Simulations ◽

Learning Approach ◽

Machine Learning Approach ◽

Chemical Reaction Mechanisms

Combining reinforcement learning (RL) and molecular dynamics (MD) simulations, we propose a machine-learning approach, called RL‡, to automatically unravel chemical reaction mechanisms. In RL‡, locating the transition state of a...

Download Full-text

Summit: Benchmarking Machine Learning Methods for Reaction Optimisation

10.26434/chemrxiv.12939806.v2 ◽

2020 ◽

Author(s):

Kobi Felton ◽

Jan Rittig ◽

Alexei Lapkin

Keyword(s):

Machine Learning ◽

Experimental Data ◽

Open Source ◽

Chemical Reaction ◽

Learning Strategy ◽

New Products ◽

Fine Chemicals ◽

Optimal Solutions ◽

Machine Learning Methods ◽

Open Source Framework

In the fine chemicals industry, reaction screening and optimisation are essential to development of new products. However, this screening can be extremely time and labor intensive, especially when intuition is used. Machine learning offers a solution through iterative suggestions of new experiments based on past experimental data, but knowing which machine learning strategy to apply in a particular case is still difficult. Here, we develop chemically-motivated virtual benchmarks for reaction optimisation and compare several strategies on these benchmarks. The benchmarks and strategies are encompassed in an open source framework named Summit. The results of our tests show that Bayesian optimisation strategies perform very well across the types of problems faced in chemical reaction optimisation, while many strategies commonly used in reaction optimisation fail to find optimal solutions.

Download Full-text