Evolutionary computation

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

An evolutionary optimizer is an optimization algorithm that searches candidate configurations through population-based variation and selection.

Evolutionary computation (EC) is a family of AI methods that simulates biological or natural processes to solve complex problems. It searches for effective individuals or populations in an artificial environment, using natural selection as a guiding model rather than calculus-based optimization alone. section 1.1.1 section 4.5

An EC problem needs a representation of an individual, a way for individuals to change, and a fitness measure. Fitness quantifies how well an individual survives, performs, or solves the target problem; it drives the search toward stronger candidate solutions. section 1.1.1

EC supplies practical search strategies for machine-learning work:

• Optimize deep-learning systems across stages from data collection through validation, providing an AutoML-oriented toolset for difficult configuration choices. section 1.1 section 1.5
• Tune hyperparameters by encoding settings as vectors, converting vector values into bounded parameter ranges, training a model, and using loss as fitness. section 5.4.2
• Approximate continuous, discontinuous, and complex functions with differential evolution, including problems not limited to smooth or convex solutions. section 4.5
• Search CNN architectures when manually choosing convolutional layers, kernels, and pooling settings requires trial and error. section 7.1.2

Using particle search for hyperparameter optimization

Represent each candidate as a vector. Map its elements onto hyperparameter ranges such as middle-layer size and learning rate, keep fixed values such as batch size and epochs, train the resulting model, and return its loss as fitness. A DEAP toolbox generates a population, evaluates particles, records their fitness and decoded settings, retains the best candidate, and updates the population over iterations. section 5.4.2

EC is useful alongside deep learning rather than only as a replacement for it. Evolutionary deep learning combines evolutionary methods with DL to automate tuning, data and feature work, model choices, and architecture exploration. It can reduce manual experimentation, but evolutionary search adds computational cost and may require distributed computing for serious projects. section 1.5.6 section 12.5.4

Figure 1.4 Applying EC (EDL) to the AI/ML model development workflow for DL

Use the workflow figure to locate evolutionary strategies across the AI/ML development process and relate search targets to data, model, and validation work. section 1.5

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

How deep it goes

Defines populations and fitness, then develops simulations, differential evolution, and DEAP-based particle-search implementations without formal proofs.

Examples

Uses Conway’s Game of Life, continuous and discontinuous function approximation, CNN layer experiments, and DEAP particle-based hyperparameter tuning.

What it assumes

Assumes familiarity with deep learning while teaching evolutionary-computation fundamentals from the ground up.

What you can do after

Apply it via DEAP to encode, evaluate, and iteratively improve deep-learning hyperparameter configurations.

How it connects

Connects deep learning, AutoML, hyperparameter optimization, genetic algorithms, differential evolution, particle swarm optimization, and neuroevolution.

liveBookmanning.com

Read Evolutionary Deep Learning: Genetic algorithms and neural networks on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Optimization Algorithms: AI techniques for design, planning, and control problemsEvolutionary computation is a family of computational methods inspired by biological evolution that iteratively improves candidate solutions through variation and selection.
• 

  liveBookmanning.com
  Grokking AI Algorithms, Second EditionA population-based algorithmic technique that iteratively improves candidate solutions through variation and selection inspired by biological evolution.

Compare how 3 books explain evolutionary computation →

Related concepts

• genetic algorithmincludes · A genetic algorithm is an evolutionary optimization method that improves a population of candidate solutions through fitness-based selection and variation, often including recombination and mutation.
• evolutionuses · Evolution is the cumulative change in the characteristics, structure, or behavior of a population or system across successive generations or states.
• artificial intelligencekind of · Artificial intelligence is the field concerned with creating computational systems that perform tasks associated with human intelligence, such as learning, reasoning, perception, or decision-making.
• particle swarm optimizationincludes · Particle swarm optimization is a population-based optimization method in which candidate solutions adjust their positions using individual and collective search information.
• life simulationincludes · Life simulation is a computational approach that models observed natural life processes such as movement, interaction, growth, and death.
• binary-coded genetic algorithmincludes · A binary-coded genetic algorithm is a genetic algorithm whose candidate solutions use binary encoding.

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.