Machine learning em química orgânica

Abstract

The search for intelligent computational systems capable of solving problems that are traditionally reserved for the human mind are a long-standing crusade. Several attempts to solve problems such as structural identification and synthesis of organic molecules using those toolboxes started in the 1970s, but the low capacity of available computational power and the lack of appropriate algorithms at the time were severe limitations, rendering many of these projects unfeasible. Currently, with a continuous increase in the processing capacity and with an enormous amount of chemical information accumulated in public and commercial databases, the interest in developing these systems has resurged. Several papers that have been published show that using machine learning algorithms it is possible to create programs capable of automatically generate synthetic paths for complex molecules of industrial and academic interest and also optimize reactions in an efficient and autonomous way.