Timur Madzhidov , Assistant Professor, Department of Organic and Medicinal Chemistry, Kazan Federal University
Do you find it a challenge selecting optimal reaction conditions in the development of a synthesis strategy, or are you developing different approaches such as neural network based methods for building predictors to aid reaction condition optimization?
If yes, then join us on October 7 for a 45 minute webinar that will cover the workflow that starts with input of raw data from Reaxys database and ends up with the model that suggests several possible reaction conditions for desired synthesis outcomes.
The workflow implies a reaction data curation procedure that includes reaction structure curation as well as standardization of reaction condition information.
Timur Madzhidov , Associate Professor & Chair of Organic Chemistry at Kazan Federal University, will discuss:
• The nearest neighbor and neural network-based approaches applied for reaction condition prediction in small and quite homogeneous hydrogenation reactions subset.
• The problems of scaling of mentioned approaches to a larger and more diverse dataset
• AI-based approaches that can be applied to a diverse database of reactions. The approaches are validated in retrospective and prospective scenarios and show quality superior to existing ones.
Part of the Big Data in Chemistry webinar series:
Showcasing the work of top class chemistry researchers who have used Elsevier’s reaction and substance data from Reaxys database as input to their current research programs for algorithm development, predictive analytics and/or data visualisation for delivering actionable insights to accelerate synthesis workflows.