Nevergrad: A Python library for performing derivative-free ML optimization

Re:Nevergrad: A Python library for performing derivative-fre... by lostmsu on Thursday December 20, 2018 @07:26PM (#18730515)

There was a recent paper from Uber, that GA works well for weights, so I wouldn't drop that area right away.

[ Parent | on HN ]

Re:Nevergrad: A Python library for performing derivative-fre... by oteytaud on Thursday December 20, 2018 @11:29PM (#18731412)

Sure GA can be great for weights as well - but mainly when gradient is unreliable. I would not use Nevergrad for training the weights of a convolutional network for image classification for example; whereas I use Nevergrad for WorldModels.

[ Parent | on HN ]

Re:Nevergrad: A Python library for performing derivative-fre... by lostmsu on Thursday December 27, 2018 @07:39PM (#18773938)

Doesn't the model Uber used begin with a bunch of convolutional layer sets, since it processes raw images?

[ Parent | on HN ]

Re:Nevergrad: A Python library for performing derivative-fre... by mi_lk on Thursday December 20, 2018 @09:57PM (#18731128)

What’s GA here?

[ Parent | on HN ]

Re:Nevergrad: A Python library for performing derivative-fre... by oteytaud on Thursday December 20, 2018 @11:27PM (#18731407)

GA stands for genetic algorithms.

[ Parent | on HN ]