liveBook · Manning

Species

As explained in Evolutionary Deep Learning: Genetic algorithms and neural networks

A species is a group or subset of evolutionary individuals grouped by similar features for population-level evaluation and management.

A species groups evolutionary individuals—here, neural-network genomes with similar features—so NEAT can evaluate and manage related networks as populations. section 10.6

NEAT uses compatibility to control which genomes share a species. A compatibility threshold determines the number of groups, while stagnation tracking identifies species that stop improving. section 10.5.1 section 10.6

Tune speciation experimentally by changing:

Circles classification

During NEAT evolution on circles, early generations track three species. Species can later become stagnant, be culled, and make room for new groups within the fixed population. section 10.5.1 section 10.6

Listing 10.19 EDL_10_5_NEAT_Speciation.ipynb: NEAT configuration options
[NEAT]
fitness_criterion     = max
fitness_threshold     = .85                 #1
pop_size              = 100
reset_on_extinction   = 1
 
[DefaultGenome]
num_inputs              = 2
num_hidden              = 25                #2
num_outputs             = 1
initial_connection      = partial_direct 0.5
feed_forward            = True
compatibility_disjoint_coefficient    = 1.0
compatibility_weight_coefficient      = 0.6
conn_add_prob           = 0.02              #3
conn_delete_prob        = 0.02              #3
node_add_prob           = 0.02              #4
node_delete_prob        = 0.02              #4

Inspect the NEAT configuration that combines compatibility-based grouping with mutation probabilities for evolving network structure. section 10.5.1

How Evolutionary Deep Learning: Genetic algorithms and neural networks covers it

How deep it goes
Explains grouping, compatibility thresholds, stagnation checks, and extinction, but stops short of deriving the compatibility distance.
Examples
Demonstrates species with the circles classification problem, configuration listings, training figures, and parameter-change exercises.
What you can do after
Operate/configure NEAT speciation by adjusting compatibility, stagnation, and population parameters and inspecting evolutionary results.
How it connects
Connects species with NEAT, evolutionary optimization, network topology, fitness evaluation, and stagnation.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Evolutionary Deep Learning: Genetic algorithms and neural networks. The section links point to the parts of the book it draws on.