CWSF 2026 Candidate

EvoNash Scientific Dashboard

Investigating Adaptive Mutation Rates in Genetic Neural Networks

Joel deFouw
Junior - Grade 8 | Digital Technology / Computing & Information Systems
Experiments
Convergence Rate
Speed Improvement
Statistical Power
Significance

1. Abstract

150-word summary of the research

EvoNash: Accelerating Convergence to Nash Equilibrium

Investigating Adaptive Mutation Rates in Genetic Neural Networks

This experiment investigates the efficiency of evolutionary algorithms in finding Nash Equilibrium in a competitive multi-agent environment (Tag). We compare a standard static mutation rate against a novel adaptive mutation strategy where the mutation rate scales inversely with an agent's fitness score. The hypothesis is that adaptive mutation—mimicking biological 'stress-induced mutagenesis'—will allow low-fitness populations to explore the solution space aggressively while high-fitness populations exploit their successful strategies, resulting in significantly faster convergence to a stable strategy (Nash Equilibrium).

2. Problem Statement

Introduction and background research

The Challenge

Deep Reinforcement Learning (DRL) is computationally expensive and hyperparameter-sensitive. Simple evolutionary algorithms are robust but often slow to converge because a fixed mutation rate is inefficient: too high disrupts good policies, too low causes stagnation. finding the optimal balance is difficult.

Background Knowledge

Nash Equilibrium

A stable state in a game where no player can improve their outcome by unilaterally changing their strategy.

Adaptive Mutation

Dynamically adjusting the rate of genetic change based on performance (fitness). High stress (low fitness) = High mutation.

Policy Entropy

A measure of the randomness of an agent's actions. High entropy = exploration; Low entropy = exploitation/convergence.

3. Hypothesis

The testable prediction guiding this research

IF

the mutation rate of a neural network is dynamically scaled inversely to its fitness score,

THEN

the population will reach a state of Policy Entropy stability (Nash Equilibrium) in fewer generations than a control group with a static mutation rate,

BECAUSE

this mechanism mimics biological 'stress-induced mutagenesis,' allowing poor-performing agents to explore the solution space rapidly while high-performing agents preserve their successful traits, balancing exploration and exploitation more efficiently.

4. Variables

The experimental design controls all factors except the independent variable

Independent Variable

The factor that is intentionally changed

Mutation Strategy
Control (Static) vs. Adaptive (Fitness-Scaled)
Adaptive
Generation Count
Number of evolutionary generations
Continuous

Dependent Variables

The outcomes being measured

Convergence Velocity
Generations to reach Nash Equilibrium (Policy Entropy < 0.001)
Policy Entropy
Measure of strategy randomness (Shannon Entropy)
Entropy Variance
Stability of population strategies
Fitness Score
Relative skill level (Self-play performance)

Controlled Variables

Factors kept constant to ensure validity

Population Size1000
Fixed number of agents per generation
Selection PressureTop 20%
Tournament size / truncation ratio
Network Architecture24-64-4
Input/Hidden/Output layers
Simulation Ticks750
Ticks per generation
Random Seed PairingsMatched
Identical seeds for Control/Experimental pairs