Understanding Names with Neural Networks - Session 1

Explore the challenges and solutions in matching names across languages and writing systems in this webinar on neural network applications. Delve into the limitations of traditional methods like edit distance and Hidden Markov Models, and discover how deep neural networks can significantly improve accuracy in English/Japanese name matching. Learn about sequence-to-sequence models, Long Short-Term Memory cells, and Convolutional Neural Networks as applied to natural language processing. Gain insights into the complexities of transliteration, scoring methods, and processing times for name pairs. Understand the critical importance of accurate name matching in various consumer and governmental domains.

Syllabus

Intro
ABOUT BASIS TECHNOLOGY
Rosette Capabilities
Names are a Challenge
Task: Name Matching
Searching Names in Watch Lists
Name Matching Algorithms
Step One: Modeling Sequences of Characters
Step Two: Modeling Transliterations
Issues with HMM-Based Name Matching
How Would You Transliterate a Name?
What does an HMM Actually Do?
The Antidote: Sequence-to-Sequence (seq2seq)
Neural Network of Choice: Long Short-Term Memory (LSTM) Cells
Step One: Learning to Transliterate with seq2seq
Step Two: Running the Transliterator in Reverse to Score
How Can We Produce a Score?
Processing Time on Name Pairs (seconds)
Convolutional Neural Net (CNN)
CNN in Natural Language Processing
What Does This Tell Us?