A Odena, I Goodfellow - stat, 2018 - wcventure.github.io
Abstract Machine learning models are notoriously difficult to interpret and debug. This is particularly true of neural networks. In this work, we introduce automated software testing …
A Odena, C Olsson, DG Andersen, I Goodfellow - proceedings.mlr.press
Neural networks are difficult to interpret and debug. We introduce testing techniques for neural networks that can discover errors occurring only for rare inputs. Specifically, we …
The underlying software implementation of these operations may contain many branching statements but many of these are based on the size of the matrix and thus the architecture of …
A Odena, C Olsson, DG Andersen, I Goodfellow - openreview.net
Neural networks are difficult to interpret and debug. We introduce testing techniques for neural networks that can discover errors occurring only for rare inputs. Specifically, we …
A Odena, I Goodfellow - stat, 2018 - szu-se.github.io
Abstract Machine learning models are notoriously difficult to interpret and debug. This is particularly true of neural networks. In this work, we introduce automated software testing …
A Odena, I Goodfellow - arXiv e-prints, 2018 - ui.adsabs.harvard.edu
Abstract Machine learning models are notoriously difficult to interpret and debug. This is particularly true of neural networks. In this work, we introduce automated software testing …
A Odena, I Goodfellow - arXiv preprint arXiv:1807.10875, 2018 - arxiv.org
Machine learning models are notoriously difficult to interpret and debug. This is particularly true of neural networks. In this work, we introduce automated software testing techniques for …