GOOD.ood_algorithms.algorithms.DANN

Implementation of the DANN algorithm from “Domain-Adversarial Training of Neural Networks” paper

Classes

DANN(config)

Implementation of the DANN algorithm from "Domain-Adversarial Training of Neural Networks" paper
class GOOD.ood_algorithms.algorithms.DANN.DANN(config: Union[CommonArgs, Munch])[source]

Bases: BaseOODAlg

Implementation of the DANN algorithm from “Domain-Adversarial Training of Neural Networks” paper

Args:

config (Union[CommonArgs, Munch]): munchified dictionary of args (config.model.model_level, config.metric.cross_entropy_with_logit(), config.ood.ood_param)

loss_postprocess(loss: Tensor, data: Batch, mask: Tensor, config: Union[CommonArgs, Munch], **kwargs) → Tensor[source]

Process loss based on DANN algorithm

Parameters

  • loss (Tensor) – base loss between model predictions and input labels

  • data (Batch) – input data

  • mask (Tensor) – NAN masks for data formats

  • config (Union[CommonArgs, Munch]) – munchified dictionary of args (config.model.model_level, config.metric.cross_entropy_with_logit(), config.ood.ood_param)

config = munchify({model: {model_level: str('graph')},
                       metric: {cross_entropy_with_logit()},
                       ood: {ood_param: float(0.1)}
                       })
Returns (Tensor):

loss based on DANN algorithm

output_postprocess(model_output: Tensor, **kwargs) → Tensor[source]

Process the raw output of model; get domain classifier predictions

Parameters

model_output (Tensor) – model raw output

Returns (Tensor):

model raw predictions