Out of order: using AI to decode the bizarre personality of water

Researchers at the University of Osaka have used machine learning to develop a systematic framework for structural descriptors in supercooled water. A neural network model was used to quantitatively assess the performance of 16 structural descriptors in differentiating between two states of water, a high-density liquid and a low-density liquid. The framework could be used to further our understanding of the anomalous behavior of water.

Raccontata daeurekalert.org