Scientists from the University of Bristol’s Intangible Realities Laboratory (IRL) and ETH Zurich have used virtual reality (VR) and artificial intelligence (AI) to learn details of reactive potential energy. Next, these details will then be transferred over to neural nets in order for them to function more effectively
Evidently, the researchers describe how advanced interaction and visualization frameworks using VR will enable humans to train machine learning algorithms and accelerate scientific discovery.
The team uses a real-time framework called Narupa. In short, Narupa is a state-of-the-art open source project that they have been developing over the last few years. It combines real-time atomistic physics simulations with commodity VR hardware. In other words, the framework can essentially carry out quantum mechanic calculations whenever needed.
Royal Society researcher Dr. David Glowacki, IRL head across Bristol’s Department of Computer Science and School of Chemistry, states, “immersive tools like VR provide an efficient means for humans to express high-level scientific and design insight. Certainly, this work represents the first time that VR framework has been used to generate data for training a neural network.” This cutting edge technology has never been seen before. Therefore, this technology can lead to great improvements in neural net training.
In conclusion, Professor Markus Reiher from ETH states, “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.”
Read the edited Science Daily article here
Reference:
Silvia Amabilino, Lars A. Bratholm, Simon J. Bennie, Alain C. Vaucher, Markus Reiher, David R. Glowacki. Training Neural Nets to Learn Reactive Potential Energy Surfaces using Interactive Quantum Chemistry in Virtual Reality. The Journal of Physical Chemistry A, 2019; DOI: 10.1021/acs.jpca.9b01006