ERRATUM article

Front. Comput. Neurosci., 07 July 2020

Volume 14 - 2020 | https://doi.org/10.3389/fncom.2020.00067

Erratum: Contextual Integration in Cortical and Convolutional Neural Networks

  • Frontiers Media SA, Lausanne, Switzerland

Due to a production error, there was a mistake in Table 1 as published. The bold entries indicating the highest accuracy for each case were un-bolded erroneously except for column SNP, 0.1. The correct Table 1 with bold values in each column appears below.

Table 1

ModelsOriginalAWGNSPN
–0.10.20.30.40.50.10.20.30.40.5
CNN98.7198.6198.2196.8892.0381.7897.2892.0180.8565.2948.28
CNNEx97.2597.1796.8395.8693.3488.2496.0693.4587.9777.9963.04
CNNEx (avg)98.7198.5898.1596.8391.8981.9097.3392.1180.7964.8747.94
CNNEx (lr)97.2597.1896.8395.8793.3788.2996.0893.4987.9978.0063.10
CNNEx (s)97.4097.3897.0096.1393.8088.8496.3493.9388.4478.4663.47

Model accuracy (%) on the MNIST dataset.

We separate results for the original images and the two types of noise perturbations by columns (AWGN, additive white gaussian noise; SPN, salt-and-pepper noise). The results for the baseline model (CNN) and the model with lateral connections (CNNEx) are shown in the first two rows. The third row [CNNEx(avg)] shows results comparable to the baseline model (CNN) when we replaced the weights in Equation (5) with a uniform distribution of weights (w = 1/NT where NT is the total number of lateral connections in each layer). The last two rows, lr and s correspond to models with just the low-rank and just the sparse component, respectively of the inhibitory lateral connections. Including lateral connections seems to improve performance with increasing noise. Using only the sparse inhibitory component also increases performance, suggesting a regularizing effect. All reported values are averages over 10 random initializations.

Bold values represent highest accuracy for each case.

The publisher apologizes for this mistake. The original article has been updated.

Summary

Keywords

contextual modulation, convolutional neuronal network, canonical cortical microcircuit, inhibitory cell types, extraclassical receptive field, lateral connectivity, natural scene statistics, Bayesian inference

Citation

Frontiers Production Office (2020) Erratum: Contextual Integration in Cortical and Convolutional Neural Networks. Front. Comput. Neurosci. 14:67. doi: 10.3389/fncom.2020.00067

Received

08 June 2020

Accepted

09 June 2020

Published

07 July 2020

Approved by

Frontiers Editorial Office, Frontiers Media SA, Switzerland

Volume

14 - 2020

Updates

Copyright

*Correspondence: Frontiers Production Office

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Cite article

Copy to clipboard


Export citation file


Share article

Article metrics