New neural architecture THEIA learns complete Kleene three-valued logic

Researchers have developed THEIA, a novel modular neural architecture comprising 2.75 million parameters, capable of learning the complete Kleene three-valued logic (K3) truth table directly from data. While Transformer and MLP baselines can also achieve high accuracy on K3 rules, THEIA demonstrates unique properties in asymmetric propagation of uncertainty and a reliability spectrum under discretized training. This architecture shows superior generalization on complex sequential tasks compared to standard MLPs and Transformers, maintaining accuracy over extended compositions where others collapse. AI

IMPACT Presents a novel approach to neural logic learning, potentially advancing AI's reasoning capabilities in uncertain environments.

RANK_REASON Academic paper detailing a new neural architecture and its capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Augustus Haoyang Li
• Kleene three-valued logic
• THEIA
• Transformer

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →