Learning to Compose Neural Networks for Question Answering
This addresses the challenge of building flexible QA systems for multiple domains, though it appears incremental as it builds on modular neural network concepts.
The paper tackles the problem of question answering across images and knowledge bases by introducing a model that automatically assembles neural networks from composable modules using natural language strings, achieving state-of-the-art results on benchmark datasets.
We describe a question answering model that applies to both images and structured knowledge bases. The model uses natural language strings to automatically assemble neural networks from a collection of composable modules. Parameters for these modules are learned jointly with network-assembly parameters via reinforcement learning, with only (world, question, answer) triples as supervision. Our approach, which we term a dynamic neural model network, achieves state-of-the-art results on benchmark datasets in both visual and structured domains.