Creating an Ensemble of Document Classifiers for Production at Internet Archive

Internet Archive has a collection of 150M PDFs and wants to improve the curation of the content by identifying which PDFs are research works.
This webinar is an experience report of what it is like to use transfer learning to make an ensemble of classifiers (FastText, Inception, BERT), define a REST interface, and put it into production using Tensorflow-serving.

When

Friday November 15th, 2019 10:00 AM PST / 12 PM CST / 1 PM EST.

Links