Series Overview

This series explores neuroevolution—the application of evolutionary algorithms to automatically design neural network architectures. Rather than manually crafting network topologies, these techniques let evolution discover optimal structures and connection patterns through biological principles of variation, selection, and inheritance.

What You’ll Learn

The Journey

NEAT (NeuroEvolution of Augmenting Topologies) introduces the foundational algorithm that revolutionized neuroevolution. Learn how NEAT solves the competing conventions problem, protects innovation through speciation, and incrementally evolves network complexity while maintaining meaningful genetic representation.

HyperNEAT addresses NEAT’s scalability limitations through indirect encoding and geometric principles. Discover how Compositional Pattern Producing Networks (CPPNs) can generate connection patterns based on spatial relationships, enabling evolution of brain-scale networks with biological regularities like symmetry and modularity.

Technical Innovation

The series covers breakthrough concepts that remain relevant today:

Modern Applications

These techniques influenced contemporary approaches in:

Biological Inspiration

Neuroevolution draws from fundamental biological principles:

Perfect for researchers in evolutionary computation, neural architecture search, or anyone interested in bio-inspired approaches to AI. These techniques offer alternatives to gradient-based optimization and insights into how complex structures might emerge through evolutionary processes.