More info from your documents: AI handwriting recognition and automatic parsing (sponsored by AXA)
Your company has a large amount of data locked into thousands or millions of scanned paper documents. You’d like to extract and analyze it, but you first have to prove that your algorithm works and brings business value.
Pick one document type associated with a high-value use case, then follow Ciprian Tomoiaga’s advice on how human insight will greatly improve a state-of-the-art deep learning model. Through careful error analysis, Ciprian walks you through the steps of turning an existing model to the specificities of your image data—even when this data is not available. He applies these insights to handwriting recognition in challenging forms and makes it available in open source so you can also benefit from it.
This session is sponsored by AXA.
What you'll learn
- Learn how to apply a deep learning model to handwriting recognition
Ciprian Tomoiaga is a computer vision engineer at AXA, where he builds models to assist the company’s 105 million clients. Previously, he spent one year at CERN, where he developed software for designing its future accelerators; then he pursued a master’s degree at École Polytechnique Fédérale de Lausanne, Switzerland, where he aided the time machine project to uncover data from the past and studied how languages evolve; this led him into handwriting recognition, which he applied to AXA’s challenging documents. He contributed to his team’s new approach for handwritten field extraction and recognition and their work was published in ICDAR’19 conference. As an undergraduate at the University of Manchester, Ciprian chaired student programming societies and contributed to the creation of now-renowned hackathons. He is still a hacker at heart, and this motivates him to be part of REV, AXA’s innovation department that puts data and AI to good use for AXA’s clients.
For conference registration information and customer service
For more information on community discounts and trade opportunities with O’Reilly conferences
For information on exhibiting or sponsoring a conference
For media/analyst press inquires