Accelerating NeurASP with vectorization and caching
This work addresses a scalability bottleneck in neurosymbolic AI for researchers using NeurASP, but the improvements are incremental as they optimize an existing framework without introducing new paradigms.
The authors improved the computational performance of the NeurASP framework by vectorizing, batching, and caching intermediate computations, achieving speedups of multiple orders of magnitude on larger tasks. They also introduced a new playing card dataset to test the enhanced learning.
Neurosymbolic AI combines neural networks with symbolic programs to create robust and explainable predictions. One such framework is NeurASP, which trains a neural network to predict concepts and reasons over them using rules written in answer set programming (ASP) to solve downstream tasks. Crucially, labels are only provided for the downstream prediction produced by the symbolic rules, not for the latent concepts themselves.Backpropagation through the non-differentiable ASP component requires expensive probability and gradient calculations, which has hindered scalability to more sophisticated tasks.In this paper, we address the current limitations of NeurASP by improving its computational performance through vectorization, batch processing and caching of intermediate computations during training. We compare computation speeds between the original and our new implementation of NeurASP and report speedups of multiple orders of magnitude for larger tasks. To this end, we propose a new dataset of difficult tasks involving playing cards, which we use to test the capabilities of NeurASP's enhanced learning function.