Collection:
Recently, a special type of data poisoning (DP) attack against deep neural network (DNN) classifiers, known as a backdoor, was proposed. These attacks do not seek to degrade classification accuracy, but rather to have the classifier learn to classify to a
- IEEE MemberUS $11.00
- Society MemberUS $0.00
- IEEE Student MemberUS $11.00
- Non-IEEE MemberUS $15.00
Videos in this product
Revealing Backdoors, Post-Training, In Dnn Classifiers Via Novel Inference On Optimized Perturbations Inducing Group Misclassification
Recently, a special type of data poisoning (DP) attack against deep neural network (DNN) classifiers, known as a backdoor, was proposed. These attacks do not seek to degrade classification accuracy, but rather to have the classifier learn to classify to a