Deep Pyramidal Residual Networks with Separated Stochastic Depth

By Yoshihiro Yamada, Masakazu Iwamura, Koichi Kise. 2016

28 citations recorded by the source corpus.

On general object recognition, Deep Convolutional Neural Networks (DCNNs) achieve high accuracy. In particular, ResNet and its improvements have broken the lowest error rate records. In this paper, we propose a method to successfully combine two ResNet improvements, ResDrop and PyramidNet. We confirmed that the proposed network outperformed the conventional methods; on CIFAR-100, the proposed network achieved an error rate of 16.18% in contrast to PiramidNet achieving that of 18.29% and ResNeXt 17.31%.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

A Comprehensive Survey of Neural Architecture Search (2020, neutral): “9 PyramidSepDrop + ShakeDrop[133] CoRR16 2.”
A Survey on Neural Architecture Search (2019, methodology): “A Wide-ResNet (Zagoruyko and Komodakis, 2016), three versions of Shake-Shake (Gastaldi, 2017) and a PyramidNet (Yamada et al., 2016) with ShakeDrop (Yamada et al., 2018) are trained with different augmentation strategies as detailed in the following.”
Shakedrop Regularization for Deep Residual Learning (2018, methodology): “proposed for ResNet, and also applied to PyramidNet [34].”
A Survey of Model Compression and Acceleration for Deep Neural Networks (2017, methodology): “Following this direction, thew work in [62] proposed a pyramidal residual networks with stochastic”
Model Compression and Acceleration for Deep Neural Networks: The Principles, Progress, and Challenges (2018, methodology): “Following this direction, the work in [62] proposed a pyramidal residual network with stochastic depth.”
A survey of the recent architectures of deep convolutional neural networks (2019, methodology): “Different researchers used a hybrid of the already proposed architectures to improve deep CNN performance (Huang et al. 2016a; Szegedy et al. 2016a; Targ et al. 2016; Yamada et al. 2016; Kuen et al. 2017; Lv et al. 2019).”
Born Again Neural Networks (2018, methodology): “It is only beaten by Yamada et al. (2018) who use a pyramidal ResNet trained for 1800 epochs with a combination of shake-shake (Gastaldi, 2017), pyramid-drop (Yamada et al., 2016) and cut-out regularization (DeVries & Taylor, 2017).”