Dilated DenseNets for Relational Reasoning
This addresses the computational inefficiency in relational reasoning for AI systems, though it appears incremental as it builds on existing DenseNet and dilated convolution techniques.
The paper tackled the problem of deep neural networks struggling with relational reasoning by introducing a DenseNet with dilated convolutions, which achieved strong performance on the Sort-of-CLEVR dataset without needing a costly relational module.
Despite their impressive performance in many tasks, deep neural networks often struggle at relational reasoning. This has recently been remedied with the introduction of a plug-in relational module that considers relations between pairs of objects. Unfortunately, this is combinatorially expensive. In this extended abstract, we show that a DenseNet incorporating dilated convolutions excels at relational reasoning on the Sort-of-CLEVR dataset, allowing us to forgo this relational module and its associated expense.