Deep Convolutional Neural Network Design Patterns

Abstract

Recent research in the deep learning field has produced a plethora of new architectures. At the same time, a growing number of groups are applying deep learning to new applications. Some of these groups are likely to be composed of inexperienced deep learning practitioners who are baffled by the dizzying array of architecture choices and therefore opt to use an older architecture (i.e., Alexnet). Here we attempt to bridge this gap by mining the collective knowledge contained in recent deep learning research to discover underlying principles for designing neural network architectures. In addition, we describe several architectural innovations, including Fractal of FractalNet network, Stagewise Boosting Networks, and Taylor Series Networks (our Caffe code and prototxt files is available at https://github.com/iPhysicist/CNNDesignPatterns). We hope others are inspired to build on our preliminary work.

Submitted as a conference paper at ICLR 2017

References in corpus (23)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Striving for Simplicity: The All Convolutional Net
• Densely Connected Convolutional Networks
• Wide Residual Networks
• Resnet in Resnet: Generalizing Residual Architectures
• FractalNet: Ultra-Deep Neural Networks without Residuals
• Residual Networks Behave Like Ensembles of Relatively Shallow Networks
• Residual Networks of Residual Networks: Multilevel Residual Networks
• Xception: Deep Learning with Depthwise Separable Convolutions
• Data Programming: Creating Large Training Sets, Quickly
• Deep Networks with Stochastic Depth
• Identity Mappings in Deep Residual Networks
• Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
• Swapout: Learning an ensemble of deep architectures
• Noisy Activation Functions
• Competitive Multi-scale Convolution
• Multi-Residual Networks: Improving the Speed and Accuracy of Residual Networks
• Deep Pyramidal Residual Networks
• Streaming Normalization: Towards Simpler and More Biologically-plausible Normalizations for Online and Recurrent Learning
• The Unreasonable Effectiveness of Noisy Data for Fine-Grained Recognition
• Understanding Locally Competitive Networks
• Convolutional Residual Memory Networks
• Frankenstein: Learning Deep Face Representations using Small Data

Cited by in corpus (8)

• A Comprehensive Analysis of Deep Regression
• Learning physical properties of liquid crystals with deep convolutional neural networks
• Deep Convolutional Neural Networks with Merge-and-Run Mappings
• Revealing Fine Structures of the Retinal Receptive Field by Deep Learning Networks
• Revealing structure components of the retina by deep learning networks
• Towards the Next Generation of Retinal Neuroprosthesis: Visual Computation with Spikes
• Neuron Segmentation Using Deep Complete Bipartite Networks
• Assisting human experts in the interpretation of their visual process: A case study on assessing copper surface adhesive potency