Paper Overview
Field: Machine Learning Author: Wael AbdAlmageed arXiv: 2507.18392
Abstract (translated)
In many reasoning problems, the premises are not observed as discrete symbols, but must be inferred from high-dimensional inputs. Furthermore, the predicate vocabulary, argument structure, and trusted evidence are supplied by a Knowledge Graph (KG) or rule definitions. Classical neuro-symbolic pipelines have a discrete interface between perception and deduction.
SoftReason is a neuro-soft-symbolic architecture for differentiable deductive reasoning over latent perceptual facts and knowledge-provided predicates. It removes the gradient gap by representing the deductive state as a local soft interpretation tensor over candidate constants and predicates.
- Perception proposes probabilistic base facts.
- KG triples enter as high-confidence soft evidence.
- Every query anchor, predicate choice, and closure update remains differentiable.
- Paper: https://arxiv.org/abs/2507.18392
The core innovation is a learned differentiable lifting of the immediate-consequence operator: it uses predicate definition embeddings and latent composition channels to form soft predicate mixtures, aggregates all possible witnesses, proposes query-conditioned head facts, and updates interpretations through a monotonic probabilistic OR.
The framework is instantiated on knowledge-aware visual question answering (KVQA), demonstrating end-to-end perceptual grounding, KG evidence injection, and differentiable deductive closure within a single trainable architecture.