🧬

Dynamic Graph Topology

Unlike traditional deep learning with fixed layers, NeuronGraph uses a dynamic directed graph structure where neurons and synapses can be added or removed at runtime.

⚡

Local Learning Rules

Implements biologically-inspired learning through STDP (Spike-Timing Dependent Plasticity), Hebbian learning, and Homeostasis without global backpropagation.

🚀

WebGPU Accelerated

High-performance GPU computation using WebGPU API. Runs natively in modern browsers with parallel processing of thousands of neurons and synapses.