Test error at 95%TPR. :return: robust test error :rtype: float

FPR at 95%TPR. :return: robust test error :rtype: float

Test error at 95%TPR. :return: robust test error :rtype: float

Test error at 95%TPR. :return: robust test error :rtype: float

Test error at 95%TPR. :return: robust test error :rtype: float

FPR at 95%TPR. :return: robust test error :rtype: float

Test error at 95%TPR. :return: robust test error :rtype: float

See https://stackoverflow.com/questions/1176136/convert-string-to-python-class-object?utm_medium=organic&utm_source=google_rich_qa&utm_campaign=g

Get exponential scheduler. Note that the resulting optimizer's step function is called after each batch! :param optimizer: optimizer

Add figure. :param tag: tag :type tag: str :param figure: test :type figure: matplotlib.pyplot.figure

Add figure. :param tag: tag :type tag: str

Add histogram data. :param tag: tag :type tag: str :param values: values :type values: torch.Tensor or numpy

Add histogram data. :param tag: tag :type tag: str

Add image. :param tag: tag :type tag: str :param img_tensor: image :type img_tensor: torch.Tensor or numpy.n

Add image. :param tag: tag :type tag: str

Add images. :param tag: tag :type tag: str :param img_tensor: images :type img_tensor: torch.Tensor or numpy

Add images. :param tag: tag :type tag: str

Gets mask in image shape given mask coordinates :param mask_coords: mask coordinates for the batch of masks :type mask_coord

Add scalar value. :param tag: tag for scalar :type tag: str

Add scalar values. :param tag: tag for scalar :type tag: str :param tag_scalar_dict: values :type tag_scalar

Add scalar values. :param tag: tag for scalar :type tag: str

Add images. :param tag: tag :type tag: str :param text_string: test :type text_string: str :param gl

Add images. :param tag: tag :type tag: str

Guard to check that no variables have been defined yet.

Get hostname. :return: hostname :rtype: str

Get layer names. :return: layer names :rtype: [str]

Linear transition rule. :param perturbations: perturbations :type perturbations: torch.autograd.Variable :param norm: norm :type

Load model files and perturbation (i.e., adversarial example) files. :param training_configs: training configurations obtained from module()

Load adversarial probabilities from perturbation files. :param perturbations_files: perturbation files as from load() :type perturba

Generate path to log directory. :param directory: directory :type directory: str

Generate path to log file. :param directory: directory :type directory: str :param name: name of file :type name: str :retur

Main.

Main LaTeX table.

Main LaTeX table.

Generate the main results table.

Loss. :param logits: predicted classes :type logits: torch.autograd.Variable :param targets: target classes :type targets: torch

Loss. :param logits: predicted classes :type logits: torch.autograd.Variable :param targets: target classes :type targets: torch

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Load module and training as well as attack configuration. :param config_module: config module such as config.mnist :type config_module:

Normalization. :param gradients: gradients :type gradients: torch.autograd.Variable

Normalization. :param gradients: gradients :type gradients: torch.autograd.Variable

Normalization. :param gradients: gradients :type gradients: torch.autograd.Variable

Convert class labels to one-hot vectors. :param classes: classes as B x 1 :type classes: torch.autograd.Variable or torch.Tensor :pa

Convert an array of numbers to an array of one-hot vectors. :param array: classes to convert :type array: numpy.ndarray :param N: nu

Create partial while preserving __name__ and __doc__. :param f: function :type f: callable :param args: arguments :type args: di

PID. :return: PID :rtype: int

Power transition rule. :param perturbations: perturbations :type perturbations: torch.autograd.Variable :param norm: norm :type

Add a layer. :param name: layer name :type name: str :param layer: layer :type layer: torch.nn.Module

Compute the orthogonal projection of the input tensor (as vector) onto the L_ord epsilon-ball. **Assumes the first dimension to be batch dim

Compute the orthogonal projection of the input tensor (as vector) onto the L_ord epsilon-ball. **Assumes the first dimension to be batch dim

Taken from https://gist.github.com/daien/1272551/edd95a6154106f8e28209a1c7964623ef8397246. Compute the Euclidean projection on a positive si

Compute the orthogonal projection of the input tensor (as vector) onto the L_ord epsilon-ball. **Assumes the first dimension to be batch dim

Generates n mask coordinates randomly from allowed locations :param n: number of masks to generate, defaults to 1 :type n: i

Random test images. :return: filepath :rtype: str

Random test labels. :return: filepath :rtype: str

Random train images. :return: filepath :rtype: str

Random train labels. :return: filepath :rtype: str

Raw Cifar10 training directory. :return: fielpath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Train images of MNSIT. :return: filepath :rtype: str

Raw SVHN training directory. :return: fielpath :rtype: str

Raw SVHN training directory. :return: fielpath :rtype: str