Adaptive importance sampling to accelerate training of a neural probabilistic language model
{{output}}
Previous work on statistical language modeling has shown that it is possible to train a feedforward neural network to approximate probabilities over sequences of words, resulting in significant error reduction when compared to standard baseline models based on... ...
