A new method for dialog state tracking using memory networks.
problem Accurately estimate the current dialog status from noisy observations.
method End-to-End Memory Network (MemN2N) for hidden state variable inference.
result The proposed tracker gives encouraging results on DSTC-2 dataset.