2023/04/01 by Awni Altabaa, Altabaa, Awni, Taylor Webb +6 · 1 voice · 2 citations
Computer Science · Mathematics · Psychology · #Advanced Text Analysis Techniques #Child and Animal Learning Development #Cognitive Science and Mapping #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #cs.LG #stat.ML
paper · pdf · doi:10.48550/arxiv.2304.00195
openalex publication_date 2023/04/01 · arxiv published 2023/04/01 · openalex created_date 2023/04/06 · arxiv updated 2024/04/12 · openalex updated_date 2026/07/28
An extension of Transformers is proposed that enables explicit relational reasoning through a novel module called the Abstractor. At the core of the Abstractor is a variant of attention called relational cross-attention. The approach is motivated by an architectural inductive bias for relational learning that disentangles relational information from object-level features. This enables explicit relational reasoning, supporting abstraction and generalization from limited data. The Abstractor is first evaluated on simple discriminative relational tasks and compared to existing relational architectures. Next, the Abstractor is evaluated on purely relational sequence-to-sequence tasks, where dramatic improvements are seen in sample efficiency compared to standard Transformers. Finally, Abstractors are evaluated on a collection of tasks based on mathematical problem solving, where consistent improvements in performance and sample efficiency are observed.