As the use of Deep Neural Networks (DNNs) be- comes pervasive, their vulnerability to adversarial attacks and limitations in handling unseen classes poses significant challenges. The state-of-the-art offers discrete solutions aimed to tackle individual issues covering specific adversarial attack scenarios, classification or evolving learning. However, real-world systems need to be able to detect and recover from a wide range of adversarial attacks without sacrificing classification accuracy and to flexibly act in unseen scenarios. In this paper, UNICAD, is proposed as a novel framework that integrates a variety of techniques to provide an adaptive solution.
For the targeted image classification, UNICAD is able to provide accurate image classification while still handling un- seen scenarios by detecting unseen classes and detecting and recovering adversarially attacked inputs. This has been achieved by leveraging Prototype and Similarity-based DNNs, along with denoising autoencoders. Our experiments performed on the CIFAR-10 dataset highlight UNICAD’s effectiveness in adver- sarial mitigation and unseen class classification, outperforming traditional models.