From Eigenmodes to Proofs: Integrating Graph Spectral Operators with Symbolic Interpretable Reasoning
Andrew Kiruluta and Priscilla Burity · 2025
Read From Eigenmodes to Proofs: Integrating Graph Spectral Operators with Symbolic Interpretable Reasoning by Andrew Kiruluta and Priscilla Burity (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-07017.txt.
Description
Logical reasoning carried out in the frequency domain of a knowledge graph is proposed by Andrew Kiruluta and one co-author in a 2025 AI paper. Their framework, Spectral NSR, encodes logical rules as templates in that same domain and reasons there directly. It uses the Laplacian eigenstructure, the graph equivalent of a musical scale.
Add-ons include learning the graph and its basis as training proceeds, sharper selective filters, and mixtures of spectral experts for modular work. Proof-guided training and uncertainty scores for well-judged confidence are also included. Coupling to large language models and robustness against attacks are also covered. On reasoning benchmarks such as ProofWriter and CLUTRR, the authors report better accuracy, faster inference, and clearer reasoning than transformer, message-passing, and neuro-symbolic baselines. Analyses show the model's decisions align closely with symbolic proof forms, and transfer tests support adaptation to new domains.
Catalog record
| CREATOR | Andrew Kiruluta and Priscilla Burity |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-07017