Multiresolution Recurrent Neural Networks: An Application to Dialogue Response Generation

We introduce the multiresolution recurrent neural network, which extends the sequence-to-sequence framework to model natural language generation as two parallel discrete stochastic processes: a sequence of high-level coarse tokens, and a sequence of natural language tokens. There are many ways to estimate or learn the high-level coarse tokens, but we argue that a simple extraction procedure is sufficient to capture a wealth of high-level discourse semantics... (read more)

PDF Abstract

Datasets


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Dialogue Generation Twitter Dialogue (Noun) MrRNN Act.-Ent. Precision 4.82 # 1
Recall 5.22 # 1
F1 4.63 # 1
Dialogue Generation Twitter Dialogue (Tense) MrRNN Act.-Ent. Accuracy 34.48% # 1
Dialogue Generation Ubuntu Dialogue (Activity) MrRNN Act.-Ent. Precision 16.84 # 1
Recall 9.72 # 1
F1 11.43 # 1
Dialogue Generation Ubuntu Dialogue (Cmd) MrRNN Act.-Ent. Accuracy 95.04% # 1
Dialogue Generation Ubuntu Dialogue (Entity) MrRNN Act.-Ent. Precision 4.91 # 1
Recall 3.36 # 1
F1 3.72 # 1
Dialogue Generation Ubuntu Dialogue (Tense) MrRNN Act.-Ent. Accuracy 29.01% # 1

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet