Talking with Reddit: Exploring limitations of recurrent neural network conversational systems.
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Microsoft Research Cambridge Talks Admins.
Recently, there has been great interest in building conversational systems based on recurrent neural networks. If trained on a large-scale corpus of conversation-like data, these models can produce surprisingly fluent responses when interacting with a human. I will discuss my implementation of a neural network conversational agent and describe the output of this model after training on Reddit conversations. While this model is able to produce plausible responses in a variety of settings, I will concentrate on the cases where the model fails, and briefly describe attempts to incorporate information from knowledge-bases to mitigate these failures.Abstract not available
This talk is part of the Microsoft Research Cambridge, public talks series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|