The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Identifying New Materials with NLP with Anubhav Jain - TWIML Talk #291
Aug 15, 2019
Join Anubhav Jain, a Staff Scientist at Lawrence Berkeley National Lab and leader of the Hacking Materials Research Group, as he dives into the intersection of materials science and natural language processing. He discusses his groundbreaking paper on using unsupervised word embeddings to analyze scientific literature, enabling advanced material discovery. Anubhav highlights innovative predictive methods and the significant role of NLP in identifying new functional materials, as well as the exciting potential of validating predictions with real experimental data.
AI Snips
Chapters
Transcript
Episode notes
Anubhav's Path to Materials Science
- Anubhav Jain's interest in materials science stems from its impact on technological advancements.
- He uses computer simulations and data mining to accelerate the discovery of new materials with interesting properties.
Tiered Screening for New Materials
- Simulated data is now used in a tiered screening process for new materials.
- Machine learning predictions, trained on simulated data, identify candidates, followed by simulations and experiments.
Mining Scientific Literature with NLP
- Scientific literature contains vast knowledge, but current methods of information extraction are limited.
- Anubhav Jain's system uses NLP to read and synthesize information from research articles, enabling testable predictions.