Research on Oral English Dialogue Understanding Based on Deep Learning

Oral English dialogue is a crucial part of system that enables computer to “understand” the input language as human does, so performance closely related oral understanding. In task-based human-machine systems, external knowledge bases can provide machine with valid information beyond training data, helping model better perform comprehension task. this paper, we propose deep recurrent neural network based on feature fusion, which directly stacks multiple nodes at single time node deepen complexity nonlinear transformations. The fusion structure applied ATIS dataset for experiments, and experimental results demonstrate RNN further improve effectiveness GRU unit obtain among different units.