Deep Reinforcement Learning for Goal oriented Dialogue Systems
Despite recent advancements in NLP, Dialogue Policy managers in most deployed Dialogue systems are still hand-coded and require a considerable amount of human effort.
These manually coded dialogue policies tend to be static and deteriorate over time due to a lack of adaptation to changes in the environment like new products and changing user behavior.
In this talk, Rajesh will demonstrate how Deep Reinforcement Learning can be used to automatically extract optimal dialogue policies from unannotated conversation logs.
Distinguished Data Scientist at PayPal
Rajesh Munavalli is a Distinguished Data Scientist at PayPal. His primary area of research includes Deep Reinforcement Learning, Natural Language Processing, and Deep Learning in the domain of Customer Service applications.
In the past, he has also worked and lead various front-end Fraud and Risk models for online payment processing.