Compositional Generalization with Tree Stack Memory Units
This addresses a key limitation in neural networks for tasks requiring compositional reasoning, such as mathematical problem-solving, though it is an incremental improvement over existing memory-augmented architectures.
The paper tackles the problem of compositional generalization, where standard neural networks struggle with zero-shot generalization to novel concept compositions, by proposing Tree Stack Memory Units (Tree-SMU), a recursive neural network with differentiable stack memory, and demonstrates strong empirical results on mathematical reasoning benchmarks, outperforming baselines like Transformers and Tree-LSTMs.
We study compositional generalization, viz., the problem of zero-shot generalization to novel compositions of concepts in a domain. Standard neural networks fail to a large extent on compositional learning. We propose Tree Stack Memory Units (Tree-SMU) to enable strong compositional generalization. Tree-SMU is a recursive neural network with Stack Memory Units (\SMU s), a novel memory augmented neural network whose memory has a differentiable stack structure. Each SMU in the tree architecture learns to read from its stack and to write to it by combining the stacks and states of its children through gating. The stack helps capture long-range dependencies in the problem domain, thereby enabling compositional generalization. Additionally, the stack also preserves the ordering of each node's descendants, thereby retaining locality on the tree. We demonstrate strong empirical results on two mathematical reasoning benchmarks. We use four compositionality tests to assess the generalization performance of Tree-SMU and show that it enables accurate compositional generalization compared to strong baselines such as Transformers and Tree-LSTMs.