Re:Neuroevolution of Self-Interpretable Agents
by hardmaru on Thursday March 19, 2020 @05:36AM
(#22625901)
In terms of sample efficiency, it's not - we're looking at 10x difference. But in terms of final performance, after training, it is (and even gets higher scores) for many tasks.
But here, we are not looking at the problem of sample efficiency, but rather at performance at generalization, and we find that evolution allows us to design and train systems that don't need to be differentiable (i.e. have many discrete operations) that may help with generalization.
But here, we are not looking at the problem of sample efficiency, but rather at performance at generalization, and we find that evolution allows us to design and train systems that don't need to be differentiable (i.e. have many discrete operations) that may help with generalization.