Loading...
Close
XCurveXCurve
Sidebar

Search tutorials

Get Started

Documentation

Example Projects

XCurve.Losses.OpenAUC

We specify the model, the loss function, and the optimizer. To generate open-set samples by the manifold mixup technique, a wrapper model, which will return the manifold, is defined in XCurve.OpenAUC.models.wrapper_classes. Besides, we define the loss StandardOpenAUCLoss, whose essential parameters are explained below: 

  CLASS StandardOpenAUCLoss(loss_close, alpha=2, lambda=0.05)     [SOURCE]
Parameters:
loss_close (nn.Module)

This parameter specifies the loss function for close-set samples;

alpha (float)

This parameter specifies the beta distribution that generate the ratio for manifold mixup.

lambd (float)

This parameter spcifies the hyper-parameter to balance the close-set loss and the open-set loss.

You can refer to the [Link] for an example.