Utku Ozbulak

Utku Ozbulak

Research Professor at Ghent University Global Campus

About me

I'm currently working as a Research Professor at Ghent University Global Campus (GUGC) and an Adjunct Professor at George Mason University Korea (GMUK), currently teaching Data Visualization at GMUK and Bioinformatics at GUGC.

My interests lie in developing AI-driven solutions that enhance clinical decision-making, improve interpretability, and ensure reliability in real-world medical and biomedical settings. These days, I'm working on surgical video understanding with video object segmentation models such as SAM2, AI-assisted characterization of sub-visible particles in biopharmaceuticals (classification, clustering, and segmentation), and other topics involving spurious correlations, robust predictions, and interpretable AI. If you're working on cool projects within my areas of interest that I can contribute to, feel free to reach out!


Here’s a brief overview of my career, including education, work history, research focus (past and present), publications, and repositories.

Summary

Current Positions:

  • Research Professor at Ghent University Global Campus, South Korea
  • Adjunct Professor at George Mason University Korea, South Korea

Education:

  • PhD in Computer Science Engineering, Ghent University, Belgium
  • MSc in Data Science, University of Southampton, United Kingdom
  • BSc in Computer Engineering, Yasar University, Turkey
Work Experience
2025 - Current: Adjunct Professor, George Mason Korea, South Korea

2023 - Current: Research Professor, Ghent University Global Campus, South Korea

2022 - 2023: Postdoctoral Fellow, Ghent University Global Campus, South Korea

2017 - 2022: AI Researcher, Ghent University Global Campus, South Korea

2015 - 2016: SAP Business Intelligence Consultant, The Coca Cola Company, Turkey

2014 - 2015: SAP Business Intelligence Consultant, Turkish Airlines, Turkey
Research Expertise

(Med) Medical and biomedical imaging

(XAI) Trustworthy and explainable AI

(SecureAI) AI security and safety

(Bio) Bioinformatics and genomics

(SSL) Image-based self-supervised learning

Publications
First or Corresponding Author Publications

(SSL-XAI) Token-Based Detection of Spurious Correlations in Vision Transformers
2026, Transactions on Machine Learning Research

(Med) Token-based Fidelity Scoring for Trustworthy Vision Transformer Interpretations in Medical Imaging
2026, International Journal of Computer Assisted Radiology and Surgery, Springer

(Med) Improved Sub-visible Particle Classification in Flow Imaging Microscopy via Generative AI-based Image Synthesis
2026, Journal of Pharmaceutical Sciences, Elsevier

(SecureAI) Exact Feature Collisions in Neural Networks
2026, Scientific Reports, Nature Publishing

(Med) Revisiting the Evaluation Bias Introduced by Frame Sampling Strategies in Surgical Video Segmentation Using SAM2
2025, MICCAI - FAMI Workshop

(Med) SpurBreast: A Curated Dataset for Investigating Spurious Correlations in Real-world Breast MRI Classification
2025, MICCAI - Main Track - Early Accept

(Med) When Tracking Fails: Analyzing Failure Modes of SAM2 for Point-Based Tracking in Surgical Videos
2025, MICCAI - COLAS Workshop

(Med) Towards Affordable Tumor Segmentation and Visualization for 3D Breast MRI Using SAM2
2025, MICCAI - Deep Breath Workshop

(Bio) Assessing the Reliability of Point Mutation as Data Augmentation for Deep Learning with Genomic Data
2024, BMC Bioinformatics

(SSL) Self-Supervised Benchmark Lottery on ImageNet: Do Marginal Improvements Translate to Improvements on Similar Datasets?
2024, IEEE IJCNN, oral presentation

(Med) Color Flow Imaging Microscopy Improves Identification of Stress Sources of Protein Aggregates in Biopharmaceuticals
2024, MICCAI - MOVI Workshop

(Med) Exploring Patient Data Requirements in Training Effective AI Models for MRI-based Breast Cancer Classification
2024, MICCAI - Deep Breath Workshop

(Med-XAI-SSL) Evaluating Visual Explanations of Attention Maps for Transformer-Based Medical Imaging
2024, MICCAI - iMIMIC Workshop

(Med-XAI-SSL) Identifying Critical Tokens for Accurate Predictions in Transformer-based Medical Imaging Models
2024, MICCAI - MLMI Workshop

(Bio-XAI) Utilizing Mutations to Evaluate Interpretability of Neural Networks on Genomic Data
2023, NeurIPS - LMRL Workshop

(Bio-XAI) Mutate and Observe: Utilizing Deep Neural Networks to Investigate the Impact of Mutations on Translation Initiation
2023, Bioinformatics, Oxford Press

(SSL) Know Your Self-supervised Learning: A Survey on Image-based Generative and Discriminative Training
2023, Transactions on Machine Learning Research

(SecureAI-XAI) Evaluating Adversarial Attacks on ImageNet: A Reality Check on Misclassification Classes
2022, NeurIPS - Workshop on ImageNet: Past, Present, and Future

(SecureAI) Selection of Source Images Heavily Influences the Effectiveness of Adversarial Attacks
2021, BMVC, oral presentation

(SecureAI-XAI) Investigating the Significance of Adversarial Attacks and Their Relation to Interpretability for Radar-based Human Activity Recognition Systems
2021, Computer Vision and Image Understanding, Elsevier

(SecureAI) Regional Image Perturbation Reduces Lp Norms of Adversarial Examples While Maintaining Model-to-model Transferability
2020, ICML - UDL Workshop

(SecureAI) Perturbation Analysis of Gradient-based Adversarial Attacks
2020, Pattern Recognition Letters, Elsevier

(SecureAI-Med) Impact of Adversarial Examples on Deep Learning Models for Biomedical Image Segmentation
2019, MICCAI, poster presentation

(SecureAI) Not All Adversarial Examples Require a Complex Defense: Identifying Over-optimized Adversarial Examples with IQR-based Logit Thresholding
2019, IEEE IJCNN, oral presentation

(SecureAI) How the Softmax Output is Misleading for Evaluating the Strength of Adversarial Examples
2018, NeurIPS - SecML Workshop

Other Publications

(Med) Predicting Agitation Stability of Monoclonal Antibodies during Developability Assessment
2026, Molecular Pharmaceutics

(Med) Balancing Redundancy and Diversity: An In-Depth Analysis of Active Learning for Laparoscopic Video Segmentation
2025, MICCAI - DEMI Workshop

(Med) One Patient's Annotation is Another One's Initialization: Towards Zero-Shot Surgical Video Segmentation with Cross-Patient Initialization
2025, Arxiv

(Bio) BRCA Gene Mutations in dbSNP: A Visual Exploration of Genetic Variants
2023, Arxiv

(Med) Tryp: A Dataset of Microscopy Images of Unstained Thick Blood Smears for Trypanosome Detection
2023, Scientific Data, Nature Publishing, Q1 SCIE

(Bio-Med) Automatic Detection of Trypanosomosis in Thick Blood Smears Using Image Pre-processing and Deep Learning
2023, International Conference on Intelligent Human Computer Interaction (IHCI)

Repositories