Understanding Names with Neural Networks

Matching names across languages and writing systems is a critical issue in a variety of consumer and governmental domains. Historically, computers have attempted to solve this problem with ad-hoc methods such as edit distance, sound indexing, and Hidden Markov Models, but these have a variety of practical limitations in this problem space, which we will explore. To address these issues, we present our research and development team’s work on doing English/Japanese name matching using deep neural networks, which provides a substantial boost in accuracy.

A 40-minute presentation will be followed by a Q&A session.