Learning Visual Reasoning Without Strong Priors
This addresses the challenge of learning multi-step reasoning from images and language without strong priors, showing that specialized architectures may not be necessary, which is incremental but impactful for multi-modal AI.
The paper tackles the problem of artificial visual reasoning by proposing a general-purpose Conditional Batch Normalization approach, achieving a 2.4% error rate on the CLEVR benchmark, outperforming other methods.
Achieving artificial visual reasoning - the ability to answer image-related questions which require a multi-step, high-level process - is an important step towards artificial general intelligence. This multi-modal task requires learning a question-dependent, structured reasoning process over images from language. Standard deep learning approaches tend to exploit biases in the data rather than learn this underlying structure, while leading methods learn to visually reason successfully but are hand-crafted for reasoning. We show that a general-purpose, Conditional Batch Normalization approach achieves state-of-the-art results on the CLEVR Visual Reasoning benchmark with a 2.4% error rate. We outperform the next best end-to-end method (4.5%) and even methods that use extra supervision (3.1%). We probe our model to shed light on how it reasons, showing it has learned a question-dependent, multi-step process. Previous work has operated under the assumption that visual reasoning calls for a specialized architecture, but we show that a general architecture with proper conditioning can learn to visually reason effectively.