🐦 Flappy Bird Neuroevolution

Watch AI learn to play using genetic algorithms with real-time neural network visualization

🧠 Best Bird's Brain Activity

Output (Jump Decision): 0.00
Positive Weight
Negative Weight
High Activation
Generation
1
Alive
0
Best Score
0
Current Best
0
Population Size:
1 50
20 birds per generation
Simulation Speed:

How it works:

Neural Network Architecture: Each bird has a brain with 5 input neurons (bird Y position, velocity, distance to pipe, pipe top Y, pipe bottom Y), 8 hidden neurons with sigmoid activation, and 1 output neuron (jump if > 0.5). Connections show weights (green = positive, red = negative, thickness = strength).

Genetic Algorithm: Birds that survive longer have higher fitness. They "breed" to create the next generation with 10% mutation rate on all weights and biases. Over generations, natural selection evolves better behaviors.

Single Bird Mode: Set population to 1 to watch a single bird's brain evolve across generations. Each generation, the bird's neural network is mutated slightly. This lets you observe how one brain learns and adapts over time through trial and error.