Path Length Regularization is a type of regularization for generative adversarial networks that encourages good conditioning in the mapping from latent codes to images. The idea is to encourage that a fixedsize step in the latent space $\mathcal{W}$ results in a nonzero, fixedmagnitude change in the image.
We can measure the deviation from this ideal empirically by stepping into random directions in the image space and observing the corresponding $\mathbf{w}$ gradients. These gradients should have close to an equal length regardless of $\mathbf{w}$ or the imagespace direction, indicating that the mapping from the latent space to image space is wellconditioned.
At a single $\mathbf{w} \in \mathcal{W}$ the local metric scaling properties of the generator mapping $g\left(\mathbf{w}\right) : \mathcal{W} \rightarrow \mathcal{Y}$ are captured by the Jacobian matrix $\mathbf{J_{w}} = \delta{g}\left(\mathbf{w}\right)/\delta{\mathbf{w}}$. Motivated by the desire to preserve the expected lengths of vectors regardless of the direction, we formulate the regularizer as:
$$ \mathbb{E}_{\mathbf{w},\mathbf{y} \sim \mathcal{N}\left(0, \mathbf{I}\right)} \left(\mathbf{J}^{\mathbf{T}}_{\mathbf{w}}\mathbf{y}_{2}  a\right)^{2} $$
where $y$ are random images with normally distributed pixel intensities, and $w \sim f\left(z\right)$, where $z$ are normally distributed.
To avoid explicit computation of the Jacobian matrix, we use the identity $\mathbf{J}^{\mathbf{T}}_{\mathbf{w}}\mathbf{y} = \nabla_{\mathbf{w}}\left(g\left(\mathbf{w}\right)·y\right)$, which is efficiently computable using standard backpropagation. The constant $a$ is set dynamically during optimization as the longrunning exponential moving average of the lengths $\mathbf{J}^{\mathbf{T}}_{\mathbf{w}}\mathbf{y}_{2}$, allowing the optimization to find a suitable global scale by itself.
The authors note that they find that path length regularization leads to more reliable and consistently behaving models, making architecture exploration easier. They also observe that the smoother generator is significantly easier to invert.
Source: Analyzing and Improving the Image Quality of StyleGANPaper  Code  Results  Date  Stars 

Task  Papers  Share 

Image Generation  52  17.87% 
Image Manipulation  14  4.81% 
Face Generation  13  4.47% 
Disentanglement  13  4.47% 
Diversity  11  3.78% 
Face Recognition  10  3.44% 
Translation  7  2.41% 
Conditional Image Generation  7  2.41% 
Face Swapping  6  2.06% 
Component  Type 


🤖 No Components Found  You can add them if they exist; e.g. Mask RCNN uses RoIAlign 