Not all neural network architectures are created equal, some perform much
better than others for certain tasks. But how important are the weight
parameters of a neural network compared to its architecture? In this work, we
question to what extent neural network architectures alone, without learning
any weight parameters, can encode solutions for a given task. We propose a
search method for neural network architectures that can already perform a task
without any explicit weight training. To evaluate these networks, we populate
the connections with a single shared weight parameter sampled from a uniform
random distribution, and measure the expected performance. We demonstrate that
our method can find minimal neural network architectures that can perform
several reinforcement learning tasks without weight training. On a supervised
learning domain, we find network architectures that achieve much higher than
chance accuracy on MNIST using random weights. Interactive version of this
paper at https://weightagnostic.github.io/
%0 Journal Article
%1 gaier2019weight
%A Gaier, Adam
%A Ha, David
%D 2019
%K deep-learning optimization
%T Weight Agnostic Neural Networks
%U http://arxiv.org/abs/1906.04358
%X Not all neural network architectures are created equal, some perform much
better than others for certain tasks. But how important are the weight
parameters of a neural network compared to its architecture? In this work, we
question to what extent neural network architectures alone, without learning
any weight parameters, can encode solutions for a given task. We propose a
search method for neural network architectures that can already perform a task
without any explicit weight training. To evaluate these networks, we populate
the connections with a single shared weight parameter sampled from a uniform
random distribution, and measure the expected performance. We demonstrate that
our method can find minimal neural network architectures that can perform
several reinforcement learning tasks without weight training. On a supervised
learning domain, we find network architectures that achieve much higher than
chance accuracy on MNIST using random weights. Interactive version of this
paper at https://weightagnostic.github.io/
@article{gaier2019weight,
abstract = {Not all neural network architectures are created equal, some perform much
better than others for certain tasks. But how important are the weight
parameters of a neural network compared to its architecture? In this work, we
question to what extent neural network architectures alone, without learning
any weight parameters, can encode solutions for a given task. We propose a
search method for neural network architectures that can already perform a task
without any explicit weight training. To evaluate these networks, we populate
the connections with a single shared weight parameter sampled from a uniform
random distribution, and measure the expected performance. We demonstrate that
our method can find minimal neural network architectures that can perform
several reinforcement learning tasks without weight training. On a supervised
learning domain, we find network architectures that achieve much higher than
chance accuracy on MNIST using random weights. Interactive version of this
paper at https://weightagnostic.github.io/},
added-at = {2019-06-12T17:57:47.000+0200},
author = {Gaier, Adam and Ha, David},
biburl = {https://www.bibsonomy.org/bibtex/275694d5201830304a5971084244b54b8/kirk86},
description = {[1906.04358] Weight Agnostic Neural Networks},
interhash = {0c0433e4cfe0ff86716b648658626593},
intrahash = {75694d5201830304a5971084244b54b8},
keywords = {deep-learning optimization},
note = {cite arxiv:1906.04358},
timestamp = {2019-06-12T17:57:47.000+0200},
title = {Weight Agnostic Neural Networks},
url = {http://arxiv.org/abs/1906.04358},
year = 2019
}