Neural Engine

English | 中文

Simulating intelligence through principles, not chemistry. Evolving artificial brains generation by generation using genetic algorithms.

Core Principles

1. Hebbian Learning: Neurons that fire together wire together
2. Synaptic Plasticity: Use it or lose it
3. Weighted Summation & Threshold Firing: Input signals are weighted and summed; fire if above threshold
4. Selective Connectivity: Not every neuron can connect to every other
5. Four "Re"s: Re-weighting, re-connecting, re-wiring, regenerating

No Cheating Rules

• Decisions must come purely from neuron activity
• No minimax, lookup tables, or external evaluation functions
• Every output must be traceable back to specific neurons

Project Structure

├── LICENSE                     MIT License
├── README.md                   English
├── README.zh-CN.md             中文
├── 单个神经元/
    ├── single_neuron.py        Neuron creation, trigger, state management
    ├── timing_wheel.py         Timing wheel event scheduler
    └── README.md               Database schema, state machine, event queue
└── 关系管理器/
    └── README.md               Connection management, adjacency table, assemblies

Current Stage

Stage 0: Survival instinct. 2D grid world, Agent with energy, brain-driven behavior, genetic algorithm evolution.

Tech Stack

• Python
• SQLite (neuron and connection storage)
• Timing wheel algorithm (event-driven scheduling)

License

MIT