Vedrana KRIVOKUĆA HAHN (née KRIVOKUĆA)

Short biography

As biometric recognition technologies become increasingly prevalent in our day-to-day lives, the need to protect our irreplaceable biometric data is evolving from an interesting research challenge to an urgent practical requirement. Fortunately, about two decades' worth of research into this problem already exists in the associated field of Biometric Template Protection (BTP), which strives towards the development of effective mechanisms for protecting our biometric "templates" (acquired images/signals or representative features). The importance of protecting our biometric data in light of the increasing number of applications demanding this personal information for authentication purposes, drove me to pursue a PhD on the topic of BTP for fingerprints (at The University of Auckland, in New Zealand), and I have continued to focus on BTP as my main field of research in the Biometrics Security and Privacy Group at Idiap Research Institute (in Switzerland). I work primarily on investigating different types of BTP techniques for various biometric modalities (e.g., face, fingerprints, finger veins, voice), and I am also studying new evaluation strategies that would allow us to more comprehensively assess the robustness of proposed BTP methods (particularly in a practical context). Aside from pursuing my BTP research interests, I am currently managing two large projects: the "presentation attack detection" component of the EU research project SOTERIA, which aims to produce a digital, secure, and user-friendly personal data platform for European citizens; and an industrial project investigating multimodal biometric recognition in vehicles. On top of my research and project management roles, I also teach a full biometrics course as part of Idiap's Masters in Artificial Intelligence programme, which covers the functioning and evaluation of biometric recognition systems, detailed case studies on various biometric modalities, presentation attack detection, and biometric template protection.

 

Education

  • 2006 - 2009: Bachelor of Computer Systems Engineering with First Class Honours (The University of Auckland, New Zealand)
  • 2010 - 2015: Doctor of Philosophy (PhD) on biometric template protection for fingerprints (The University of Auckland, New Zealand)

 

Publications

  • A. Komaty, V. Krivokuća Hahn, C. Ecabert and S. Marcel, 2023. "Can personalised hygienic masks be used to attack face recognition systems?". In Proceedings of the 2023 International Joint Conference on Biometrics (IJCB), Ljubljana, Slovenia (pp. 1-10), IEEE, doi: 10.1109/IJCB57857.2023.10449104.
  • W. H. Abdulla, F. Marattukalam and V. Krivokuća Hahn, 2023. "Exploring Human Biometrics: A Focus on Security Concerns and Deep Neural Networks". APSIPA Transactions on Signal and Information Processing, vol. 12. no. 1, doi: 10.1561/116.00000021.
  • H. Otroshi Shahreza, V. Krivokuća Hahn and S. Marcel, 2023. "MLP-Hash: Protecting Face Templates via Hashing of Randomized Multi-Layer Perceptron". In Proceedings of the 31st European Signal Processing Conference (EUSIPCO), Helsinki, Finland (pp. 605-609), IEEE.
  • V. Krivokuća Hahn and S. Marcel, 2023. "Biometric Template Protection for Neural-Network-based Face Recognition Systems: A Survey of Methods and Evaluation Techniques". IEEE Transactions on Information Forensics and Security, vol. 18, pp. 639-666, doi: 10.1109/TIFS.2022.3228494.
  • H. Otroshi Shahreza, C. Rathgeb, D. Osorio-Roig, V. Krivokuća Hahn, S. Marcel and C. Busch, 2022. "Hybrid Protection of Biometric Templates by Combining Homomorphic Encryption and Cancelable Biometrics". In Proceedings of the 2022 International Joint Conference on Biometrics (IJCB), Abu Dhabi, United Arab Emirates (pp. 1-10), IEEE.
  • H. Otroshi Shahreza, V. Krivokuća Hahn and S. Marcel, 2022. "Face Reconstruction from Deep Facial Embeddings using A Convolutional Neural Network". In Proceedings of the 29th IEEE International Conference on Image Processing, Bordeaux, France, IEEE.
  • V. Krivokuća Hahn and S. Marcel, 2022. "Towards Protecting Face Embeddings in Mobile Face Verification Scenarios". IEEE Transactions on Biometrics, Behavior, and Identity Science, vol. 4, no. 1, pp. 117-134, doi: 10.1109/TBIOM.2022.3140472.
  • H. Otroshi Shahreza, V. Krivokuća Hahn and S. Marcel, 2021. "On the Recognition Performance of BioHashing on state-of-the-art Face Recognition models". In 2021 IEEE International Workshop on Information Forensics and Security (WIFS) (pp. 1-6). IEEE.
  • V. Krivokuća and S. Marcel, 2020. "On the recognition performance of biohash-protected finger vein templates". In Handbook of Vascular Biometrics (pp. 465-480). Springer, Cham.
  • V. Krivokuća, M. Gomez-Barrero, S. Marcel, C. Rathgeb and C. Busch, 2020. "Towards Measuring the Amount of Discriminatory Information in Finger Vein Biometric Characteristics Using a Relative Entropy Estimator". In Handbook of Vascular Biometrics (pp.507-525). Springer. Cham.
  • V. Krivokuća and S. Marcel, 2018. "Towards quantifying the entropy of fingervein patterns across different feature extractors". In 2018 IEEE 4th international conference on identity, security, and behavior analysis (ISBA) (pp. 1-8). IEEE.
  • V. Krivokuća and W. Abdulla, 2016. "Cancellability and diversity analysis of fingerprint template protection scheme based on compact minutiae pattern". Information Security Journal: A Global Perspective, 25(1-3), pp.109-123.
  • V. Krivokuća and W. Abdulla, 2015. "Fingerprint template protection scheme based on partial minutiae patterns: a comprehensive non-invertibility analysis". International Journal of Biometrics, 7(4), pp.326-353.
  • V. Krivokuća and W. Abdulla, 2016. "Noninvertible fingerprint transforms: Categorization of design mechanisms and discussion of evaluation techniques". Information Security Journal: A Global Perspective, 25(4-6), pp.261-279.
  • V. Krivokuća and W. Abdulla, 2014, March. "Intra-class Variance Among Multiple Samples of the Same Person's Fingerprint in a Cooperative User Scenario". In International Conference on Pattern Recognition Applications and Methods (pp. 77-92). Springer, Cham.
  • V. Krivokuća and W. Abdulla, 2015. "Recognition accuracy of the new fingerprint construct based on a compact minutiae pattern". International Journal of Biometrics, 7(2), pp.170-189.
  • V. Krivokuća, W. H. Abdulla and A. Swain, 2014. "A non-invertible cancellable fingerprint construct based on compact minutiae patterns". International Journal of Biometrics, 6(2), pp.125-142.
  • V. Krivokuća, W. H. Abdulla and A. Swain, 2014. "Minutiae Persistence among Multiple Samples of the Same Person's Fingerprint in a Cooperative User Scenario". In ICPRAM (pp. 76-86).
  • V. Krivokuća, W. Abdulla and A. Swain, 2012. "A dissection of fingerprint fuzzy vault schemes". In Proceedings of the 27th Conference on Image and Vision Computing New Zealand (pp. 256-261).
  • V. Krivokuća and W. Abdulla, 2012. "Fast fingerprint alignment method based on minutiae orientation histograms". In Proceedings of the 27th Conference on Image and Vision Computing New Zealand (pp. 486-491).

 

Open-source projects

  • bob.paper.polyprotect_2021
    (Code to reproduce work in: V. Krivokuća Hahn and S. Marcel, 2022. "Towards Protecting Face Embeddings in Mobile Face Verification Scenarios". IEEE Transactions on Biometrics, Behavior, and Identity Science, vol. 4, no. 1, pp. 117-134, doi: 10.1109/TBIOM.2022.3140472.)
  • bob.chapter.fingerveins_relative_entropy
    (Code to reproduce work in: V. Krivokuća, M. Gomez-Barrero, S. Marcel, C. Rathgeb and C. Busch, 2020. "Towards Measuring the Amount of Discriminatory Information in Finger Vein Biometric Characteristics Using a Relative Entropy Estimator". In Handbook of Vascular Biometrics (pp.507-525). Springer. Cham.)
  • bob.chapter.fingerveins_biohashing
    (Code to reproduce work in: V. Krivokuća and S. Marcel, 2020. "On the recognition performance of biohash-protected finger vein templates". In Handbook of Vascular Biometrics (pp. 465-480). Springer, Cham.)
  • bob.paper.isba2018_entropy
    (Code to reproduce work in: V. Krivokuća and S. Marcel, 2018. "Towards quantifying the entropy of fingervein patterns across different feature extractors". In 2018 IEEE 4th international conference on identity, security, and behavior analysis (ISBA) (pp. 1-8). IEEE.)

 

Patents

  • European Patent EP3719679B1
    V. Krivokuća Hahn and S. Marcel, "A method for protecting biometric templates, and a system and method for verifying a speaker's identity" (Filed in 2019, granted on 09.06.2021.).

  • US Patent US11514918B2
    V. Krivokuća Hahn and S. Marcel, "Method for protecting biometric templates, and a system and method for verifying a speaker's identity" (Filed in 2020, granted on 29.11.2022.).

 

Public talks

  • Biometric Template Protection for Face Recognition Systems: A behind-the-scenes look at the Motivation, Methods, and Metrics (Tutorial presented at IJCB2022)
    As our world becomes increasingly inter-connected, the need for robust identity assurance is increasing in tandem. Biometric technologies have great potential to fulfil this need, which is why we are witnessing a rise in the number of applications adopting biometrics for authentication purposes. Applications range from personal, relatively low-stakes authentication scenarios (e.g., smartphone unlocking) to large-scale, high-stakes authentication (e.g., border crossing). The most widely-adopted biometric technology is face recognition (FR), the popularity of which is attributable to factors like common access to personal cameras (e.g., in smartphones) and claims of near-perfect recognition accuracy thanks to deep neural networks (DNNs). Although the security benefits associated with FR are numerous, there are growing privacy concerns over how our sensitive face data is being handled by the increasing number of applications demanding this personal information for identity management. One reason for this concern is that modern FR systems are based on DNNs, and we now know that the face features learned by these DNNs can be "inverted" to recover an approximation of the original face image, and they can also reveal additional information about the underlying person (e.g., gender, age, etc.). So, the face features employed in modern FR systems are rich with personally identifiable information, which may represent a threat to the privacy of the FR systems' users and the security of the system itself if the features are leaked to an adversary. Therefore, to ensure that we can take advantage of, and trust, FR technologies without jeopardising our security and privacy, we must establish effective means of protecting the employed face data. This need is reflected in recent privacy frameworks (e.g., the EU GDPR), which recognise the fact that biometric information is sensitive data that should be handled carefully to protect individuals' digital identities. Fortunately, the importance of protecting our biometric data has been known for over twenty years by the Biometric Template Protection (BTP) research community, which strives towards the development of effective mechanisms for protecting our biometric "templates" (acquired images/signals or representative features). This tutorial will present an overview of the BTP research field in the context of DNN-based FR systems. Although the tutorial will focus on FR as a case study (due to the widespread adoption of FR in practice), the presented BTP techniques and evaluation strategies are by no means limited to the face modality; indeed, most of these methods can be (and have been) applied to various biometric modalities. The tutorial will be structured in three main parts: (1) The motivation for protecting face templates, (2) Examples of the types of face BTP methods that have been proposed, and (3) A discussion of the metrics used to evaluate the robustness of these BTP methods. The aim of this tutorial is to raise awareness of the timeliness and urgency of this topic, as well as to generate more (cross-disciplinary) interest in advancing the state of the BTP research field towards effective practical deployment.
  • Biometric Template Protection for Neural-Network-based Face Recognition Systems (Talk presented at NBLAW2023)
    This talk presents an overview of the motivation behind biometric template protection (BTP) in face recognition systems, and the types of BTP methods that have been proposed for neural-network-based face recognition systems in the literature. The talk is based on the survey paper entitled "Biometric Template Protection for Neural-Network-based Face Recognition Systems: A Survey of Methods and Evaluation Techniques" (published in TIFS), as well as the IJCB2022 tutorial entitled "Biometric Template Protection for Face Recognition Systems: A behind-the-scenes look at the Motivation, Methods, and Metrics".
Link to slides

Tel: +41 27 721 7735
Office: 207-2
Contact