Noise and Error Prediction for Neural Networks

摘要

In this paper, we are studying the application of a neural network to a problem related to autonomous car driving. In particular, we are interested in the correlation between noise in the data and the average error that can be achieved by the neural network, and study the issue using several data collections. We propose a measure of the noise that can indicate issues in the collected data and can help improve the learning process. We are also interested in ways to use symmetry in the data for this particular problem and analyze how we can take advantage of it for a better training of the network.