I am reading through the NEAT paper here. On page 14 of the PDF, there is this quote about mutation:

There was an 80% chance of a genome having its connection weights mutated, in which case each weight had a 90% chance of being uniformly perturbed and a 10% chance of being assigned a new random value.

What exactly does it mean to perturb weights? What is uniform vs. nonuniform perturbation?

Is there an established method to do this? I am imagining the process as multiplying each connection weight by a random number, but I'm unfamiliar with the term.


Perturbed here means adding a small random value to the weight. That random value comes from a uniform distribution or from a gaussian (or any distribution really). Imagine just nudging the weight by a little.

It’s done to overcome the problem of local minima where models can get stuck with a good set of weights but not the best set of weights. By perturbing the weights a little, the model has a chance of finding a better set of weights. Searching for methods that deal with local minima like stochastic gradient descent will give you a better intuition.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.