Neural Network Simulation

An interactive 3D visualization of a neural network using Three.js. This simulation creates a visually stunning representation of neurons and their connections with realistic signal propagation.

Features

• Realistic Neural Structure: Neurons are organized in a central core with radiating branches
• Signal Propagation: Visual representation of neural signals traveling between neurons
• Dynamic Firing: Neurons fire spontaneously or when receiving signals
• Interactive: Click on any neuron to manually trigger it
• Performance Optimized: Uses object pooling for signal visualization
• Adjustable Parameters: Control firing rate and signal speed in real-time
• Responsive Design: Adapts to different screen sizes

Controls

• View Navigation:
  • Rotate: Click and drag
  • Zoom: Mouse wheel
  • Pan: Right-click and drag
• Neuron Interaction: Click directly on any neuron to trigger it
• Simulation Controls: Use the panel in the top-right corner to adjust parameters
  • Firing Rate: Controls the probability of spontaneous neuron firing
  • Signal Speed: Adjusts how quickly signals propagate through the network
  • Pulse Size: Change the size of signal pulses
  • Trigger Random Neurons: Manually activate random neurons
  • Auto-Rotation: Toggle automatic camera rotation

Technical Details

The simulation is built using:

• Three.js for 3D rendering
• Object Pooling for efficient particle management
• Quadratic Bezier curves for neural connections
• Ray casting for interactive neuron selection

The simulation demonstrates principles of neural networks including signal propagation and the interconnected nature of neurons in a visually engaging way.

Browser Compatibility

The simulation works best in modern browsers that support WebGL and ES6 modules.