Utku Özbulak Faculty of Engineering and Architecture > Department of Electronics and information systems Work address Technologiepark Zwijnaarde 126 9052 Zwijnaarde Email email@example.com 1 – 5 of 5 publications Show 10 5 10 15 20 50 100 250 Sort by year (new to old) Actions Filter publications Save this search Download search results Subscribe to news feed Add to list Conference Paper C1 Regional image perturbation reduces Lp norms of adversarial examples while maintaining model-to-model transferability Utku Özbulak (UGent) , Jonathan Peck (UGent) , Wesley De Neve (UGent) , Bart Goossens (UGent) , Yvan Saeys (UGent) and Arnout Van Messem (2020) the 37th International Conference on Machine Learning (ICML 2020), Proceedings. Add to list Journal Article A1 Perturbation analysis of gradient-based adversarial attacks Utku Özbulak (UGent) , Manvel Gasparyan (UGent) , Wesley De Neve (UGent) and Arnout Van Messem (2020) PATTERN RECOGNITION LETTERS. 135. p.313-320 Add to list Conference Paper P1 open access Impact of adversarial examples on deep learning models for biomedical image segmentation Utku Özbulak (UGent) , Arnout Van Messem and Wesley De Neve (UGent) (2019) MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2019, PT II. In Lecture Notes in Computer Science 11765. p.300-308 Add to list Conference Paper P1 open access Not all adversarial examples require a complex defense : identifying over-optimized adversarial examples with IQR-based logit thresholding Utku Özbulak (UGent) , Arnout Van Messem and Wesley De Neve (UGent) (2019) 2019 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN). Add to list Conference Paper C1 How the Softmax output is misleading for evaluating the strength of adversarial examples Utku Özbulak (UGent) , Wesley De Neve (UGent) and Arnout Van Messem (2018) NeurIPS 2018 : 32nd conference on neural information processing systems, Papers.