Compositional Attention Networks for Machine Reasoning

TL;DR

The paper introduces the MAC network, a neural architecture that enables explicit, interpretable, and efficient reasoning by decomposing problems into attention-based steps, achieving state-of-the-art results on visual reasoning tasks.

Contribution

It presents the MAC network, a novel differentiable architecture that improves transparency, reasoning capabilities, and data efficiency over previous models.

Findings

Achieved 98.9% accuracy on CLEVR dataset

Halved the error rate compared to previous models

Requires 5x less data for strong performance

Abstract

We present the MAC network, a novel fully differentiable neural network architecture, designed to facilitate explicit and expressive reasoning. MAC moves away from monolithic black-box neural architectures towards a design that encourages both transparency and versatility. The model approaches problems by decomposing them into a series of attention-based reasoning steps, each performed by a novel recurrent Memory, Attention, and Composition (MAC) cell that maintains a separation between control and memory. By stringing the cells together and imposing structural constraints that regulate their interaction, MAC effectively learns to perform iterative reasoning processes that are directly inferred from the data in an end-to-end approach. We demonstrate the model's strength, robustness and interpretability on the challenging CLEVR dataset for visual reasoning, achieving a new state-of-the-art 98.9% accuracy, halving the error rate of the previous best model. More importantly, we show that the model is computationally-efficient and data-efficient, in particular requiring 5x less data than existing models to achieve strong results.