Attention on Abstract Visual Reasoning
This work addresses the problem of improving performance on abstract cognitive tasks in computer vision, representing an incremental advancement by hybridizing existing methods.
The authors tackled abstract visual reasoning on image data by proposing a hybrid network architecture called Attention Relation Network (ARNe), which combines self-attention and relational reasoning, achieving an 11.28 percentage point improvement over the Wild Relation Network model on the Procedurally Generated Matrices dataset and requiring only 35% of training samples to surpass baseline accuracy.
Attention mechanisms have been boosting the performance of deep learning models on a wide range of applications, ranging from speech understanding to program induction. However, despite experiments from psychology which suggest that attention plays an essential role in visual reasoning, the full potential of attention mechanisms has so far not been explored to solve abstract cognitive tasks on image data. In this work, we propose a hybrid network architecture, grounded on self-attention and relational reasoning. We call this new model Attention Relation Network (ARNe). ARNe combines features from the recently introduced Transformer and the Wild Relation Network (WReN). We test ARNe on the Procedurally Generated Matrices (PGMs) datasets for abstract visual reasoning. ARNe excels the WReN model on this task by 11.28 ppt. Relational concepts between objects are efficiently learned demanding only 35% of the training samples to surpass reported accuracy of the base line model. Our proposed hybrid model, represents an alternative on learning abstract relations using self-attention and demonstrates that the Transformer network is also well suited for abstract visual reasoning.