A Computational Foundation for AGI
This repository contains the open-source reference implementation and the living manuscript for the Creating Intelligence project.
This project introduces a new computational theory of mind grounded in set theory and hyperdimensional computing. Unlike traditional neural networks that rely on continuous weights and matrix multiplication, this framework uses sparse binary data. It demonstrates that associative memory emerges naturally from network topologies with a combinatorially expanded layer. This opens a new route toward synthetic intelligence with human-level energy efficiency.
The zero-dependency standard-C reference implementation of the core topological associative memory algorithm.
A portable Mathematica package that implements the neural circuit components and the data flow logic.
A brief guide to building neural circuits.
A collection of circuit models demonstrating hetero-associative memory and auto-associative memory architectures.
The complete, living document exploring the mathematical, algorithmic, and neuroscientific foundations of this framework is available online.
-
For architectural discussions, conceptual questions, and sharing experiments, please join the Discussions.
-
For bug reports and specific code contributions, please use the Issue Tracker.
This repository contains two distinct categories of content, which are licensed separately:
-
The Software: Licensed under the MIT License.
-
The Manuscript & Assets: All Rights Reserved. These materials may not be reproduced, distributed, modified, translated, or commercialized without explicit, prior written permission from the author.
See the full LICENSE file for details.
Copyright © 2026 Peter Overmann. All rights reserved.