Adversarial Training: embedding adversarial perturbations into the parameter space of a neural network to build a robust system.

Shixian Wen Laurent Itti

Published in: CoRR (2019)

Keyphrases

parameter space
neural network
training algorithm
multi agent
training process
back propagation
feed forward neural networks
image space
latent space

starting points
multivariate data
backpropagation algorithm
training patterns
artificial neural networks
feed forward
parameter values
training set
multi layer perceptron

neural network training
training samples
recurrent neural networks
fuzzy logic
search space
feedforward neural networks
label noise
active learning
riemannian manifolds

neural nets
bp neural network
target function
neural network model
feature extraction
genetic algorithm
neural network is trained
machine learning