Functions — github.com/CausalLearning/HSJA

CodeHub MCPcopy Create free account

hub / github.com/CausalLearning/HSJA / functions

Functions24 in github.com/CausalLearning/HSJA

⨍ Functions 24 ◇ Types & classes 2

↓ 6 callersFunctiondecision_function

Decision function output 1 on the desired side of the boundary, 0 otherwise.

↓ 5 callersFunctionclip_image

(image, clip_min, clip_max)

↓ 5 callersFunctionresnet_layer

2D Convolution-Batch Normalization-Activation stack builder # Arguments inputs (tensor): input tensor from input image or previous layer num_fil

↓ 4 callersFunctioncompute_distance

(x_ori, x_pert, constraint = 'l2')

↓ 3 callersFunctionbinary_search_batch

Binary search to approach the boundar.

↓ 3 callersMethodpredict

(self, x, verbose=0, batch_size = 500, logits = False)

build_model.py:76

↓ 2 callersFunctionproject

(original_image, perturbed_images, alphas, params)

↓ 2 callersFunctionsplit_data

(x, y, model, num_classes = 10, split_rate = 0.8, sample_per_class = 100)

load_data.py:54

↓ 1 callersFunctionapproximate_gradient

(model, sample, num_evals, delta, params)

↓ 1 callersFunctionattack

↓ 1 callersFunctionconstruct_model_and_data

Load model and data on which the attack is carried out. Assign target classes and images for targeted attack.

↓ 1 callersFunctionconstruct_original_network

(dataset_name, model_name, train)

build_model.py:17

↓ 1 callersFunctiongeometric_progression_for_stepsize

Geometric progression to search for stepsize. Keep decreasing stepsize by half until reaching the desired side of the boundary,

↓ 1 callersFunctionhsja

Main algorithm for HopSkipJumpAttack. Inputs: model: the object that has predict method. predict outputs probability scores. clip_max: upper

↓ 1 callersFunctioninitialize

Efficient Implementation of BlendedUniformNoiseAttack in Foolbox.

↓ 1 callersFunctionphi

↓ 1 callersFunctionresnet_v2

ResNet Version 2 Model builder [b] Stacks of (1 x 1)-(3 x 3)-(1 x 1) BN-ReLU-Conv2D or also known as bottleneck layer First shortcut connection pe

↓ 1 callersFunctionselect_delta

Choose the delta at the scale of distance between x and perturbed sample.

(self, model_name, dataset_name, train = False, load = False, **kwargs)

build_model.py:53

(self, dataset_name)

load_data.py:15

Functioncreate_resnet_generator

Functionlr_schedule

Learning Rate Schedule Learning rate is scheduled to be reduced after 80, 120, 160, 180 epochs. Called automatically every epoch as part of callbac

Functionlr_schedule_cifar100

Learning Rate Schedule Learning rate is scheduled to be reduced after 80, 120, 160, 180 epochs. Called automatically every epoch as part of callbac

Functionlr_schedule_sgd