A CNN that operates on Boolean hypercubes — XOR-addressed topology, Hamming-distance kernels, end-to-end backprop in pure C++23.
A cascaded preprocessor architecture built on the hypercube substrate: an etalon-style geometric mixing stage followed by a synthetic-orbit dynamical encoding stage, feeding a hypercube CNN.
World model on a Boolean hypercube: frozen reservoir encoders and a locally connected predictor that learns how the latent code moves under an action. C++20, no dependencies.
Frozen hypercube reservoir for static high-dimensional fields (spectra, sensors, images). Short synthetic orbit, HypercubeCNN on the end state only. C++23 + Python.
Echo State Network (ESN) on a Boolean hypercube reservoir: sparse, XOR-defined connectivity and no stored adjacency. A tunable per-neuron delay line scales memory capacity super-multiplicatively. C++23 and Python SDKs.
HypercubeCNN with a hypercube preprocessor for static fields (spectra, images). A vertex and its antipode form an etalon; one reflection walk, then a thin readout. Not a reservoir. C++23.
Locally connected deep neural network on a Boolean hypercube: the cube's edges are the wiring, and every weight trains. C++ core + Python SDK.
To associate your repository with the hypercube-ai topic, visit your repo's landing page and select "manage topics."