Predicting Neural Network Accuracy from Weights

02/26/2020
by   Thomas Unterthiner, et al.
14

We study the prediction of the accuracy of a neural network given only its weights with the goal of better understanding network training and performance. To do so, we propose a formal setting which frames this task and connects to previous work in this area. We collect (and release) a large dataset of almost 80k convolutional neural networks trained on four image datasets. We demonstrate that strong predictors of accuracy exist. Moreover, they can achieve good predictions while only using simple statistics of the weights. Surprisingly, these predictors are able to rank networks trained on unobserved datasets or using different architectures.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset