Ask Me Anything: Dynamic Memory Networks for Natural Language Processing
It addresses the need for a general-purpose neural architecture in NLP, offering a unified approach for researchers and practitioners, though it is incremental in combining attention and memory mechanisms.
The paper tackles the problem of unifying diverse NLP tasks as question answering by introducing the Dynamic Memory Network (DMN), which achieves state-of-the-art results on tasks like question answering (bAbI dataset), sentiment analysis (Stanford Sentiment Treebank), and part-of-speech tagging (WSJ-PTB).
Most tasks in natural language processing can be cast into question answering (QA) problems over language input. We introduce the dynamic memory network (DMN), a neural network architecture which processes input sequences and questions, forms episodic memories, and generates relevant answers. Questions trigger an iterative attention process which allows the model to condition its attention on the inputs and the result of previous iterations. These results are then reasoned over in a hierarchical recurrent sequence model to generate answers. The DMN can be trained end-to-end and obtains state-of-the-art results on several types of tasks and datasets: question answering (Facebook's bAbI dataset), text classification for sentiment analysis (Stanford Sentiment Treebank) and sequence modeling for part-of-speech tagging (WSJ-PTB). The training for these different tasks relies exclusively on trained word vector representations and input-question-answer triplets.