Skip to main content

Module 20: EC for Neural Network Training - Evolution Meets Deep Learning

Can you train a Transformer with evolution? A CNN? An LSTM? Yes - and sometimes you should. This module covers the intersection of EC and deep learning: weight evolution, NAS, hyperparameter optimization, and hybrid gradient+evolution approaches.

Learning Objectives​

  1. Train a feedforward network using pure evolutionary weight optimization
  2. Understand evolutionary neural architecture search (NAS)
  3. Use EC for hyperparameter optimization of deep learning models
  4. Know hybrid approaches (evolutionary-guided gradient descent)
  5. Understand EvoPrompting and EC for LLM optimization

Concept Explanation​

Coming soon.

Code Examples​

Coming soon.

Exercises​

Coming soon.

Milestone Checklist​

  • Trained an NN with pure evolution
  • Understand NAS with evolutionary methods
  • Used EC for hyperparameter tuning
  • Know when evolution vs gradient descent wins for NN training

Was this page helpful?