发表机构
Chinese Institute for Brain Research, Beijing; Beijing Key Laboratory of Brain Science and Brain-Machine Interface; Sun Yat-sen University（北京脑科学与类脑研究中心; 北京脑科学与脑机接口重点实验室; 中山大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对现有方法压缩关系拓扑、缺乏可解释性的问题，提出受大脑启发的神经结构推理器NSR，通过耦合神经元群体保留结构，并行推理并实现链接预测，兼具竞争力、低训练时间和原生可解释性。
AI 中文摘要
结构推理，即识别对象与概念之间关系结构并进行推理的能力，是人类认知的标志性特征，然而现有方法往往将关系拓扑压缩为扁平嵌入，无法发现隐藏结构且缺乏可解释性。我们提出神经结构推理器（NSR），一种受大脑启发的网络，直接在耦合神经元群体的连接性和动态中保留关系结构。NSR从三种生物机制中汲取灵感：用于编码层次化知识的多层架构、实体与概念的稳定表征，以及用于输入驱动状态推理的路径整合。在查询时，NSR对候选关系结构并行化计算，并利用置信度加权分数进行链接预测。在标准知识图谱基准上，NSR取得了有竞争力的准确率，虽未在每个数据集上领先，且其训练时间低于多种神经基线。由于推理通过可读的神经元激活序列实现，NSR通过追踪中间推理步骤提供了原生可解释性。该模型还提取了潜在的关系层次和组合规则，展示了这种受大脑启发的架构作为结构推理的有效、高效且高度可解释的基底。
英文摘要
Structural reasoning, the ability to recognize and make inferences over the relational structure between objects and concepts, is a hallmark of human cognition, yet prevailing methods often collapse relational topology into flat embeddings, cannot discover hidden structure and lack interpretability. We introduce Neural Structural Reasoner (NSR), a brain-inspired network that preserves relational structure directly in the connectivity and dynamics of coupled neuronal populations. NSR draws inspiration from three biological mechanisms: multi-layered architecture for encoding hierarchical knowledge, stable representations of entity and concepts, and path integration for input-driven state inference. At query time, NSR parallelizes computation over candidate relational structures and leverages confidence-weighted scores to perform link prediction. Across standard knowledge-graph benchmarks, NSR achieves competitive accuracy without leading on every dataset, and has lower reported training times than several neural baselines. Because reasoning is implemented through sequences of human-readable neuron activations, NSR affords native interpretability by tracking intermediate inference steps. The model further extracts latent relational hierarchies and compositional rules, demonstrating the brain-inspired architecture as an effective, efficient, and highly interpretable substrate for structural reasoning.
CommentsAccepted at NeurIPS 2026
