{"author":{"name":"Augustus Haoyang Li","slug":"augustus-haoyang-li","article_count":1,"latest_published_at":"2026-04-20T12:32:35.278+00:00","profile_url":"https://platform.waiboom.ai/authors/augustus-haoyang-li","api_url":"https://platform.waiboom.ai/api/authors/augustus-haoyang-li"},"articles":[{"slug":"theia-learning-complete-kleene-three-valued-logic-in-a-pure-neural-modular-archi","title":"Modular Neural Logic: How Architecture Shapes Compositional Reasoning","url":"https://platform.waiboom.ai/article/2026/04/20/theia-learning-complete-kleene-three-valued-logic-in-a-pure-neural-modular-archi","content_type":"research_summary","summary":"Researchers present THEIA, a modular neural architecture that learns complete Kleene three-valued logic end-to-end without external symbolic solvers. The system uses four dedicated engines for arithmetic, order, set membership, and propositional logic that converge in a final logic module, achieving full K3 rule coverage on a 2M-sample dataset in under 8 minutes. Mechanistic analysis reveals that modularity enables a 'delayed verdict' strategy where upstream engines encode domain-specific variables without committing to final truth values, with the verdict emerging only at the logic boundary, a representational pattern distinct from how monolithic Transformers solve the same problem.","published_at":"2026-04-20T12:32:35.278+00:00","updated_at":"2026-05-07T02:19:52.375108+00:00","source":{"url":"https://arxiv.org/abs/2604.11284","name":"ArXiv (cs.AI)"},"featured_image":{"url":"https://www.infoworld.com/wp-content/uploads/2024/06/neural_network_by_ktsimage_gettyimages-891769302_2400x1600-100788348-orig.jpg?quality=50&strip=all&w=1024","alt":null},"categories":[{"name":"Research","slug":"research"},{"name":"AI Safety & Alignment","slug":"ai-safety-alignment"}]}]}