How do you minimize a function when you can't take derivatives? CMA-ES and PSO
What happens when you want to minimize a function, say, the error function in order to train a machine learning model, but the function has no derivatives, or they are very hard to calculate? You can use Gradient-Free optimizers. In this video, I show you two of them:
CMA-ES (Covariance matrix adaptation strategy)
PSO (Particle swarm optimization)
This video is a sequel to "What is Quantum Machine Learning"
• What is Quantum Machine Learning?
and also part of the blog post:
https://www.zapatacomputing.com/why-g...
Introduction: (0:00)
CMA-ES: (1:23)
PSO (9:17)
Conclusion: (14:00)