IBM Research | Using AI to discover new drugs and materials with limited data @ibmresearch | Uploaded April 2022 | Updated October 2024, 4 days ago.
Deep learning generative AI models are effective at coming up with novel molecules for drug and materials discovery, but they generally require large amounts of training data to learn. By treating molecules as graphs and learning the grammar of the graph, we developed a method that requires tens to hundreds of training examples, compared deep learning models that can require nearly one hundred thousand examples. This lets us generate candidates faster and more flexibly, shortening the pipeline for creating new pharmaceuticals or materials. In this talk, Dr. Jie Chen will explain this method of molecular generation, and will be joined by Dr. Thomas Asche from Evonik Industries to discuss its application in discovering new polymers.
Learn more about Jie’s work on data-efficient graph grammar to be presented at #ICLR2022 https://news.mit.edu/2022/generating-new-molecules-with-graph-grammar-0401
#deeplearning #generativemodels #chemistry #AI
Deep learning generative AI models are effective at coming up with novel molecules for drug and materials discovery, but they generally require large amounts of training data to learn. By treating molecules as graphs and learning the grammar of the graph, we developed a method that requires tens to hundreds of training examples, compared deep learning models that can require nearly one hundred thousand examples. This lets us generate candidates faster and more flexibly, shortening the pipeline for creating new pharmaceuticals or materials. In this talk, Dr. Jie Chen will explain this method of molecular generation, and will be joined by Dr. Thomas Asche from Evonik Industries to discuss its application in discovering new polymers.
Learn more about Jie’s work on data-efficient graph grammar to be presented at #ICLR2022 https://news.mit.edu/2022/generating-new-molecules-with-graph-grammar-0401
#deeplearning #generativemodels #chemistry #AI