Neural Module Networks

TL;DR

This paper introduces neural module networks that leverage linguistic question structures to dynamically compose neural modules for improved visual question answering, achieving state-of-the-art results on multiple datasets.

Contribution

It presents a novel method for constructing and training neural networks based on decomposing questions into their linguistic substructures, enabling modular and compositional reasoning.

Findings

Achieved state-of-the-art results on VQA dataset.

Performed well on a new dataset of abstract shapes.

Demonstrated the effectiveness of question decomposition for visual reasoning.

Abstract

Visual question answering is fundamentally compositional in nature---a question like "where is the dog?" shares substructure with questions like "what color is the dog?" and "where is the cat?" This paper seeks to simultaneously exploit the representational capacity of deep networks and the compositional linguistic structure of questions. We describe a procedure for constructing and learning *neural module networks*, which compose collections of jointly-trained neural "modules" into deep networks for question answering. Our approach decomposes questions into their linguistic substructures, and uses these structures to dynamically instantiate modular networks (with reusable components for recognizing dogs, classifying colors, etc.). The resulting compound networks are jointly trained. We evaluate our approach on two challenging datasets for visual question answering, achieving state-of-the-art results on both the VQA natural image dataset and a new dataset of complex questions about abstract shapes.