Researchers Use Virtual Reality and AI Algorithms to Learn the Details of Chemical Change

Researchers at the University of Bristol’s Intangible Realities Laboratory (IRL) and ETH Zurich have made use of virtual reality and artificial intelligence algorithms to study chemical change in detail.

In a cover article published in The Journal of Physical Chemistry on May 23^rd, 2019, scientists from the University of Bristol and ETH Zurich explain how sophisticated interaction and visualization frameworks based on virtual reality (VR) allow humans to train machine-learning algorithms and speed up scientific discovery.

The researchers give an account of their study developing an advanced open-source VR software framework that can perform “on-the-fly” quantum mechanics calculations.

It enables research scientists to investigate more advanced physics models of complex molecular rearrangements that involve the making and breaking of chemical bonds. This is the first-ever use of virtual reality to facilitate such a thing.

Using their interactive VR system, the researchers “taught” quantum chemistry to neural networks.

Silvia Amabilino, the lead author of the study, works between the IRL and Bristol’s Centre for Computational Chemistry.

Generating datasets to teach quantum chemistry to machines is a longstanding challenge. Our results suggest that human intuition, combined with VR, can generate high-quality training data, and thus improve machine learning models.

Silvia Amabilino, Study Lead Author, University of Bristol

According to Dr Lars Bratholm, study co-author who works between the IRL, the Centre for Computational Chemistry, and the School of Mathematics, “For most scientific computational workflows, the bottleneck is processing power. But machine learning has created a scenario where the new bottleneck is the ability to quickly generate high-quality data.”

Immersive tools like VR provide an efficient means for humans to express high-level scientific and design insight. As far as we know, this work represents the first time that a VR framework has been used to generate data for training a neural network.

Dr David Glowacki, Royal Society Research Fellow, University of Bristol

Glowacki heads the IRL across Bristol’s Department of Computer Science and School of Chemistry.

The advent of automation and machine learning across science and society has resulted in significant questions related to the sort of scientific future researchers should be consciously striving to develop over the next few decades. Descriptions of the emerging future often cast automation as the ultimate end, and at times, it is obscure where exactly the human fits in.

According to Professor Markus Reiher from ETH, “This work shows that advanced visualization and interaction frameworks like VR and AR enable humans to complement automated machine learning approaches and accelerate scientific discovery.”

“The paper offers an interesting vision for how science may evolve in the near future, where humans focus their efforts on how to effectively train machines.”

(Video credit: University of Bristol)