FractalNet

2000년 도입 · 논문 6편에서 사용

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.

출처: FractalNet: Ultra-Deep Neural Networks without Residuals

소개 논문: FractalNet: Ultra-Deep Neural Networks without Residuals