Softened Symbol Grounding for Neuro-symbolic Systems

arXiv:2403.00323v1 Announce Type: cross
Abstract: Neuro-symbolic learning generally consists of two separated worlds, i.e., neural network training and symbolic constraint solving, whose success hinges on symbol grounding, a fundamental problem in AI. This paper presents a novel, softened symbol grounding process, bridging the gap between the two worlds, and resulting in an effective and efficient neuro-symbolic learning framework. Technically, the framework features (1) modeling of symbol solution states as a Boltzmann distribution, which avoids expensive state searching and facilitates mutually …

abstract arxiv cs.ai cs.lg gap network network training neural network neuro novel paper process success systems training type