Description
FractalNet is a type of convolutional neural network that eschews residual connections in favour of a "fractal" design. They involve repeated application of a simple expansion rule to generate deep networks whose structural layouts are precisely truncated fractals. These networks contain interacting subpaths of different lengths, but do not include any pass-through or residual connections; every internal signal is transformed by a filter and nonlinearity before being seen by subsequent layers.
Papers Using This Method
Handwritten Bangla Character Recognition Using The State-of-Art Deep Convolutional Neural Networks2017-12-28CrescendoNet: A Simple Deep Convolutional Neural Network with Ensemble Behavior2017-10-30Beyond Finite Layer Neural Networks: Bridging Deep Architectures and Numerical Differential Equations2017-10-27SMASH: One-Shot Model Architecture Search through HyperNetworks2017-08-17Deep Convolutional Neural Network Design Patterns2016-11-02FractalNet: Ultra-Deep Neural Networks without Residuals2016-05-24