Neural Network
Type: Dense
Layers: 3
Neurals: 15
Current neural network layout
Input
Hidden
Output
Bird View
Bird Input/Output data visualized
Family Tree
Options
ID
Fitness
Score
Colors
Tree Graph
Run simulation and select bird from Current Gen
Current Gen
DNA
DNA updates every time new Generation is created
Fitness:
0%
Simulation
Offline
Setup
Gap
120
Birds
10
Additional Settings
Enviromental Options
Closing Rate
5000p
Pipe gap closing by 1 every time score is reaches closing rate
Number of pipes
5
How many pipes on the screen at the same time
Hardness Rate
60%
How fast pipes diverse from the center
Choice confidence
55%
How confident the bird must be to jump
Crossover Options
Pool Size
10pcs
How many top scroring birds from the generation are chosen to the mating pool
Selection Power
1
Bird score is multiplied by power so highest scoring birds are more likely to pass genes on
Mutation Options
Mutate Rate
10%
Bird mutation change
Recreate Rate
100p
If bird score is less than n, create new random bird to avoid cycle of bad generations
Attributes
Click on the screen to jump
Offline
Speed 1x
