It is difficult for researchers and scientists to test the relevance of a chemical reaction from a large host of samples. Owing to this problem, a number of commercially-relevant chemical reactions go unnoticed. To solve this problem, a new AI-based algorithm to predict chemical reactions has been developed. The use of artificial intelligence allows this platform to analyse the constituents of molecules in a reaction. The researchers have named this platform “IBM RXN for Chemistry”. The platform works similar to language translation tools such as Google Translate. It uses neural technologies to study the properties of molecules and predict reactions. The process of converting archetypes or designs to final products can be achieved with the help of this research. Furthermore, the accuracy level of the predictions is as high as 90%.
The researcher trio behind the genesis of this platform shared valuable insights from their study. They started off by permuting various possibilities of predicting reactions based on reagents, reactants, and existing conditions. The researchers further studied the likelihood of getting accurate results through the use of AI technologies. They were focused on keeping an innovative approach which further led them to use neural technologies.
Use of Ground Rules
The importance of postulates and theories within organic chemistry cannot be undermined. The researchers used the physics of chemical reactions in order to develop a neural platform for predicting reactions. A system called SMILES (simplified molecular-input line-entry system) became the basis for making the platform work. A number of patents and examples from textbooks were also used during the research.