How to build self-learning conversation systems from Pyry Takala

Most software that uses natural language to chat with humans is not intelligent. Bots usually have a number of pre-programmed answers that can only be given to very specifically worded questions. If there is input that was not anticipated by the authors, the answers become useless.
Recent advances in deep learning have made it feasible to create software that learns to reply on its own by looking at historical conversations. A learning conversation machine can answer user queries significantly more flexibly, and can constantly improve and refine its internal structure from user feedback.

In this talk, Pyry will cover the basics of how one can build self-learning conversation systems.

Bio: Pyry Takala is the co-founder of True AI, where he develops deep learning for dialogue modeling. Prior to True AI, Pyry was working on his PhD, focusing on sequence modeling, in particular applying deep learning to textual data. He has previously worked on deep learning at Amazon, and his publications have been featured for instance in MIT Technology Review and NY Times. He is also an alumni of Goldman Sachs and McKinsey.