Weight Agnostic Neural Networks
This work addresses the problem of understanding architecture importance in neural networks for researchers, but it is incremental as it builds on existing architecture search concepts.
The authors investigated whether neural network architectures alone, without trained weights, can solve tasks, and found that their search method discovered minimal architectures achieving much higher than chance accuracy on MNIST and performing reinforcement learning tasks without weight training.
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/