Attention-based Iterative Decomposition for Tensor Product Representation
This work addresses the challenge of discovering symbolic structure from unseen data for deep neural networks, representing an incremental improvement in TPR-based methods.
The paper tackled the problem of limited systematic generalization in Tensor Product Representation (TPR) models by proposing an Attention-based Iterative Decomposition (AID) module, which significantly improved performance on systematic generalization tasks compared to prior TPR-based works.
In recent research, Tensor Product Representation (TPR) is applied for the systematic generalization task of deep neural networks by learning the compositional structure of data. However, such prior works show limited performance in discovering and representing the symbolic structure from unseen test data because their decomposition to the structural representations was incomplete. In this work, we propose an Attention-based Iterative Decomposition (AID) module designed to enhance the decomposition operations for the structured representations encoded from the sequential input data with TPR. Our AID can be easily adapted to any TPR-based model and provides enhanced systematic decomposition through a competitive attention mechanism between input features and structured representations. In our experiments, AID shows effectiveness by significantly improving the performance of TPR-based prior works on the series of systematic generalization tasks. Moreover, in the quantitative and qualitative evaluations, AID produces more compositional and well-bound structural representations than other works.