Stochastic Answer Networks for Machine Reading Comprehension
This work addresses machine reading comprehension for NLP researchers, presenting an incremental improvement with a simple trick to enhance robustness.
The paper tackles the problem of multi-step reasoning in machine reading comprehension by proposing a stochastic answer network (SAN) that uses stochastic prediction dropout during training, achieving competitive results on SQuAD, Adversarial SQuAD, and MS MARCO datasets.
We propose a simple yet robust stochastic answer network (SAN) that simulates multi-step reasoning in machine reading comprehension. Compared to previous work such as ReasoNet which used reinforcement learning to determine the number of steps, the unique feature is the use of a kind of stochastic prediction dropout on the answer module (final layer) of the neural network during the training. We show that this simple trick improves robustness and achieves results competitive to the state-of-the-art on the Stanford Question Answering Dataset (SQuAD), the Adversarial SQuAD, and the Microsoft MAchine Reading COmprehension Dataset (MS MARCO).