Machine Learning for Functional Group Identification in Vibrational Spectroscopy: A Pedagogical Lab for Undergraduate Chemistry Students

Abstract

Techniques from the branch of artificial intelligence known as machine learning (ML) have been applied to a wide range of problems in chemistry. Nonetheless, there are very few examples of pedagogical activities to introduce ML to chemistry students in the chemistry education literature. Here we report a computational activity that introduces undergraduate physical chemistry students to ML in the context of vibrational spectroscopy. In the first part of the activity, students use ML binary classification algorithms to distinguish between carbonyl-containing and noncarbonyl-containing molecules on the basis of their infrared absorption spectra. In the second part of the activity, students test modifications to this basic analysis including different analysis parameters, different ML algorithms, and different test data sets. In a final extension of the activity, students implement a multiclass classification to predict whether carbonyl-containing molecules contain a ketone, a carboxylic acid, or another carbonyl group. This activity is designed to introduce students both to the basic workflow of a ML classification analysis and to some of the ways in which ML analyses can fail. We provide a comprehensive handout for the activity, including theoretical background and a detailed protocol, as well as data sets and code to implement the exercise in Python or Mathematica. This activity is designed as a standalone exercise for physical chemistry lab classes but can also be integrated with courses or modules on vibrational spectroscopy and computational chemistry. On the basis of student surveys, we conclude that this activity was successful in introducing students to applications of ML in chemistry.

eChem: A notebook exploration of quantum chemistry

“…3,8,9 The educational benefits of interactive notebooks have been realized in many fields, with examples in, e.g., bioscience and informatics, 10 and radiology physics. 11 In chemistry, specifically, notebooks have been developed on the topics of scientific computing, 12,13 analytical chemistry, 14 stochastic simulations of processes, 15 labs in physical chemistry, 16 machine learning, 17,18 molecular docking, 19 or to explain basic concepts like the hydrogen molecule, 20 the particle in a box, 21 reciprocal space, 22 and more. It has been noted that deeper insights are gained after using interactive notebooks, 21 and students have been seen to quickly adopt Jupyter notebooks also for other courses.…”

eChem: A notebook exploration of quantum chemistry

“…3,8,9 The educational benefits of interactive notebooks have been realized in several fields with examples in, e.g., bioscience and informatics, 10 and radiology physics. 11 In chemistry, specifically, notebooks have been developed on the topics of scientific computing, 12,13 analytical chemistry, 14 stochastic simulations of processes, 15 labs in physical chemistry, 16 machine learning, 17,18 molecular docking, 19 or to explain basic concepts like the hydrogen molecule, 20 the particle in a box, 21 reciprocal space, 22 and more. It has been noted that deeper insights are gained after using interactive notebooks, 21 and students have been seen to quickly adopt Jupyter notebooks also for other courses.…”

Exploring Machine Learning in Chemistry through the Classification of Spectra: An Undergraduate Project

“…ML task involves the use of binary classification algorithms alongside ML techniques to classify infrared spectra as "carbonyl" or "non-carbonyl". 13 A number of authors have approached the subject of teaching machine learning via artificial neural networks. 14,15 However, the data sets that are required for training these networks need to be very reproducible to avoid the ANNs finding patterns in the data that do not exist.…”