Publications

Papers and preprints from the group, newest first.

2026

Privacy-Preserving Detection of Rare Disease-Associated Cell Subsets via Secure Multi-Party Computation

arXiv preprint ŞS Mağara, E Havemann, D Jutz, AB Ünal, M Akgün

BibTeX

@misc{magara2026privacy,
  title         = {Privacy-Preserving Detection of Rare Disease-Associated Cell Subsets via Secure Multi-Party Computation},
  author        = {Ma{\u{g}}ara, {\c{S}}eyma Selcan and Havemann, Esther and Jutz, Debora and {\"U}nal, Ali Burak and Akg{\"u}n, Mete},
  year          = {2026},
  eprint        = {2608.20118},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.20118}
}

Empirical Evidence for Simply Connected Decision Regions in Image Classifiers

arXiv preprint A Swaminathan, M Akgün

Paper Code

Abstract

Understanding the topology of decision regions is central to explaining the inner workings of deep neural networks. Prior empirical work has provided evidence that these regions are path connected. We study a stronger topological question: whether closed loops inside a decision region can be contracted without leaving that region. To this end, we propose an iterative quad-mesh filling procedure that constructs a finite-resolution label-preserving surface bounded by a given loop and lying entirely within the same decision region.
BibTeX
```
@misc{swaminathan2026empirical,
  title         = {Empirical Evidence for Simply Connected Decision Regions in Image Classifiers},
  author        = {Swaminathan, Arjhun and Akg{\"u}n, Mete},
  year          = {2026},
  eprint        = {2605.06380},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url           = {https://arxiv.org/abs/2605.06380}
}
```
Unsupervised Multi-Source Federated Domain Adaptation under Domain Diversity through Group-Wise Discrepancy Minimization

ICML L Reichart, CA Baykara, AB Ünal, H Lee, M Akgün

Paper Code

Abstract

We propose GALA, a scalable and robust federated UMDA framework that introduces two key components: (1) an inter-group discrepancy minimization objective that approximates full pair-wise domain alignment without quadratic computation; and (2) a temperature-controlled, centroid-based weighting strategy that dynamically prioritizes source domains based on alignment with the target. To evaluate high-diversity scenarios, we introduce Digit-18, a benchmark of 18 digit datasets with varied synthetic and real-world domain shifts. Extensive experiments show that GALA matches or exceeds state-of-the-art performance on standard benchmarks and decisively outperforms prior methods in diverse multi-source settings where others fail to converge.
BibTeX
```
@inproceedings{reichart2026unsupervised,
  title     = {Unsupervised Multi-Source Federated Domain Adaptation under Domain Diversity through Group-Wise Discrepancy Minimization},
  author    = {Reichart, Larissa and Baykara, Cem Ata and {\"U}nal, Ali Burak and Lee, Harlin and Akg{\"u}n, Mete},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML)},
  year      = {2026}
}
```
Position: Trustworthy Model Context Protocol Enables Responsible Agentic AI!

ICML A Swaminathan, A Hannemann

Paper

Abstract

As a standard and interface between agents and tools, the Model Context Protocol (MCP) is a natural enforcement point—improvements to it automatically propagate to every adopting system. Examining MCP through the European Commission’s ethics guidelines for trustworthy AI, we identify fundamental shifts in how trustworthiness operates, the challenges those shifts create, and strategic intervention points where protocol-level mechanisms can drive ecosystem-wide impact. We show how MCP’s architecture already offers a foundation for trustworthiness and propose practical enhancements to strengthen it, arguing that a trustworthy MCP is essential for responsible agentic AI deployments.
BibTeX
```
@inproceedings{swaminathan2026position,
  title     = {Position: Trustworthy Model Context Protocol Enables Responsible Agentic AI!},
  author    = {Swaminathan, Arjhun and Hannemann, Anika},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML)},
  year      = {2026}
}
```
Accelerating Targeted Hard-Label Adversarial Attacks in Low-Query Black-Box Settings

SaTML A Swaminathan, M Akgün

Paper Code

Abstract

We propose Targeted Edge-informed Attack (TEA), a novel attack that utilizes edge information from the target image to carefully perturb it, thereby producing an adversarial image that is closer to the source image while still achieving the desired target classification. Our approach consistently outperforms current state-of-the-art methods across different models in low query settings (nearly 70% fewer queries are used), a scenario especially relevant in real-world applications with limited queries and black-box access. Furthermore, by efficiently generating a suitable adversarial example, TEA provides an improved target initialization for established geometry-based attacks.
BibTeX
```
@inproceedings{swaminathan2026accelerating,
  title     = {Accelerating Targeted Hard-Label Adversarial Attacks in Low-Query Black-Box Settings},
  author    = {Swaminathan, Arjhun and Akg{\"u}n, Mete},
  booktitle = {2026 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)},
  pages     = {756--767},
  year      = {2026}
}
```

A Blockchain-Based Explainable Federated Learning System for the Trustworthy Collective Defense of IoT Networks in the European Union

Studies in Big Data (Springer) I Şafak, T Frantti, M Akgün

Paper

BibTeX

@incollection{safak2026blockchain,
  title     = {A Blockchain-Based Explainable Federated Learning System for the Trustworthy Collective Defense of IoT Networks in the European Union},
  author    = {{\c{S}}afak, Ilg{\i}n and Frantti, Tapio and Akg{\"u}n, Mete},
  booktitle = {Cyber Security: Policy and Technology},
  editor    = {Lehto, Martti and Neittaanm{\"a}ki, Pekka},
  series    = {Studies in Big Data},
  volume    = {83},
  pages     = {359--384},
  publisher = {Springer Nature Switzerland},
  year      = {2026},
  doi       = {10.1007/978-3-032-08890-1_15}
}

2025

PP-GWAS: Privacy Preserving Multi-Site Genome-wide Association Studies

Nature Communications A Swaminathan, A Hannemann, AB Ünal, N Pfeifer, M Akgün

Paper Code

Abstract

We present a novel algorithm PP-GWAS designed to improve upon existing standards in terms of computational efficiency and scalability without sacrificing data privacy. This algorithm employs randomized encoding within a distributed architecture to perform stacked ridge regression on a Linear Mixed Model to ensure rigorous analysis. Experimental evaluation with real world and synthetic data indicates that PP-GWAS can achieve computational speeds twice as fast as similar state-of-the-art algorithms while using lesser computational resources, all while adhering to a robust security model that caters to an all-but-one semi-honest adversary setting.
BibTeX
```
@article{swaminathan2025ppgwas,
  title   = {{PP-GWAS}: Privacy Preserving Multi-Site Genome-wide Association Studies},
  author  = {Swaminathan, Arjhun and Hannemann, Anika and {\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  journal = {Nature Communications},
  year    = {2025},
  doi     = {10.1038/s41467-025-66771-z}
}
```
Robust Representation Learning for Privacy-Preserving Machine Learning: A Multi-Objective Autoencoder Approach

IEEE Access S Ouaari, AB Ünal, M Akgün, N Pfeifer

Paper

Abstract

Several domains increasingly rely on machine learning in their applications. The resulting heavy dependence on data has led to the emergence of various laws and regulations around data ethics and privacy and growing awareness of the need for privacy-preserving machine learning (ppML). Current ppML techniques utilize methods that are either purely based on cryptography, such as homomorphic encryption, or that introduce noise into the input, such as differential privacy. The main criticism given to those techniques is the fact that they either are too slow or they trade off a model’s performance for improved confidentiality. To address this performance reduction, we aim to leverage robust representation learning as a way of encoding our data while optimising the privacy-utility trade-off.
BibTeX
```
@article{ouaari2025robust,
  title   = {Robust Representation Learning for Privacy-Preserving Machine Learning: A Multi-Objective Autoencoder Approach},
  author  = {Ouaari, Sofiane and {\"U}nal, Ali Burak and Akg{\"u}n, Mete and Pfeifer, Nico},
  journal = {IEEE Access},
  volume  = {13},
  year    = {2025},
  doi     = {10.1109/ACCESS.2025.3603429}
}
```
PRISM: Privacy-preserving Rare Disease Analysis using Fully Homomorphic Encryption

Bioinformatics G Akkaya, N Erdoğmuş, M Akgün

Paper Code

Abstract

We present PRISM, a novel privacy-preserving framework based on fully homomorphic encryption (FHE) that facilitates rare disease variant analysis across multiple institutions without exposing sensitive genomic information. To address the challenges of centralized trust, PRISM is built upon a threshold FHE scheme. This approach decentralizes key management across participating institutions and ensures no single entity can unilaterally decrypt sensitive data. Our method filters disease-causing variants under recessive, dominant, and de novo inheritance models entirely on encrypted data.
BibTeX
```
@article{akkaya2025prism,
  title   = {{PRISM}: Privacy-preserving Rare Disease Analysis using Fully Homomorphic Encryption},
  author  = {Akkaya, G{\"u}liz and Erdo{\u{g}}mu{\c{s}}, Nesli and Akg{\"u}n, Mete},
  journal = {Bioinformatics},
  volume  = {41},
  number  = {10},
  pages   = {btaf468},
  year    = {2025},
  doi     = {10.1093/bioinformatics/btaf468}
}
```
Privacy-Preserving Federated Unsupervised Domain Adaptation with Application to Age Prediction from DNA Methylation Data

Bioinformatics CA Baykara, AB Ünal, N Pfeifer, M Akgün

Paper Code Code (FREDA-CV)

Abstract

In this work, we propose freda, a privacy-preserving federated method for unsupervised domain adaptation in regression tasks. Unlike deep learning-based FDA approaches, freda is the first method to enable the federated training of Gaussian Processes to model complex feature relationships while ensuring complete data privacy through randomized encoding and secure aggregation. This allows for effective domain adaptation without direct access to raw data, making it wellsuited for applications involving high-dimensional, heterogeneous datasets.
BibTeX
```
@article{baykara2025privacy,
  title   = {Privacy-Preserving Federated Unsupervised Domain Adaptation with Application to Age Prediction from {DNA} Methylation Data},
  author  = {Baykara, Cem Ata and {\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  journal = {Bioinformatics},
  volume  = {41},
  number  = {10},
  pages   = {btaf465},
  year    = {2025},
  doi     = {10.1093/bioinformatics/btaf465}
}
```
Secure and Efficient Logistic Regression with Secret-Sharing MPC and Differential Privacy

IEEE Access SS Magara, I Motorcu, ES Mektepli, CA Baykara, AB Ünal, M Akgün

Paper Code

Abstract

Machine learning models are being increasingly deployed in sensitive applications where data privacy and model security are of paramount importance. This paper introduces a novel privacy-preserving approach for logistic regression that integrates secret sharing-based multi-party computation (MPC) with differential privacy (DP). Our approach ensures input data confidentiality during training via a three-party MPC protocol that supports mini-batch gradient descent, reducing computational and communication overhead compared to prior MPC methods. To protect the model against membership inference and inversion attacks, we implement DP through direct gradient perturbation that scales efficiently to high-dimensional data while preserving strong (\epsilon)-differential privacy guarantees.
BibTeX
```
@article{magara2025secure,
  title   = {Secure and Efficient Logistic Regression With Secret-Sharing {MPC} and Differential Privacy},
  author  = {Ma{\u{g}}ara, {\c{S}}eyma Selcan and Motorcu, {\.I}pek and Mektepli, Emin {\c{S}}ahin and Baykara, Cem Ata and {\"U}nal, Ali Burak and Akg{\"u}n, Mete},
  journal = {IEEE Access},
  volume  = {13},
  year    = {2025},
  doi     = {10.1109/ACCESS.2025.3599456}
}
```
Enabling Privacy-preserving Model Evaluation in Federated Learning via Fully Homomorphic Encryption

IEEE TPS-ISA CA Baykara, AB Ünal, M Akgün

Paper

Abstract

Current research on federated learning primarily focuses on preserving privacy during the training phase. However, model evaluation has not been adequately addressed, despite the potential for significant privacy leaks during this phase as well. In this paper, we demonstrate that the state-of-the-art AUC computation method for federated learning systems, which utilizes differential privacy, still leaks sensitive information about the test data while also requiring a trusted central entity to perform the computations.
BibTeX
```
@inproceedings{baykara2025enabling,
  title     = {Enabling Privacy-preserving Model Evaluation in Federated Learning via Fully Homomorphic Encryption},
  author    = {Baykara, Cem Ata and {\"U}nal, Ali Burak and Akg{\"u}n, Mete},
  booktitle = {2025 IEEE 7th International Conference on Trust, Privacy and Security in Intelligent Systems, and Applications (TPS-ISA)},
  year      = {2025}
}
```
Accelerating Probabilistic Privacy-Preserving Medical Record Linkage: A Three-Party MPC Approach

Journal of Biomedical Informatics ŞS Mağara, Noah Dietrich, AB Ünal, M Akgün

Paper Code

Abstract

This work addresses the critical challenge of securely linking medical records across institutions while preserving patient privacy. Integrating data from various sources in healthcare can provide comprehensive insights into disease progression and treatment outcomes. However, strict privacy regulations often limit the ability to share this data. To tackle this, we propose a three-party Multi-Party Computation (MPC) Record Linkage method that ensures sensitive data remains confidential throughout the linkage process. By eliminating the need for Bloom filters and using bigram-based string similarity computations in a three-party framework, the method not only enhances privacy but also achieves an impressive up to 14x faster performance compared to state-of-the-art solutions. This scalable approach offers a practical solution for large-scale, privacy-preserving healthcare data integration.
BibTeX
```
@article{magara2025accelerating,
  title   = {Accelerating Probabilistic Privacy-Preserving Medical Record Linkage: A Three-Party {MPC} Approach},
  author  = {Ma{\u{g}}ara, {\c{S}}eyma Selcan and Dietrich, Noah and {\"U}nal, Ali Burak and Akg{\"u}n, Mete},
  journal = {Journal of Biomedical Informatics},
  volume  = {171},
  pages   = {104920},
  year    = {2025},
  doi     = {10.1016/j.jbi.2025.104920}
}
```
Distributed and Secure Kernel-Based Quantum Machine Learning

TMLR A Swaminathan, M Akgün

Paper Code

Abstract

While advancements in secure quantum machine learning are notable, the development of secure and distributed quantum analogues of kernel-based machine learning techniques remains underexplored. In this work, we present a novel approach for securely computing common kernels, including polynomial, radial basis function (RBF), and Laplacian kernels, when data is distributed, using quantum feature maps.
BibTeX
```
@article{swaminathan2025distributed,
  title   = {Distributed and Secure Kernel-Based Quantum Machine Learning},
  author  = {Swaminathan, Arjhun and Akg{\"u}n, Mete},
  journal = {Transactions on Machine Learning Research},
  issn    = {2835-8856},
  year    = {2025},
  url     = {https://openreview.net/forum?id=3jdI0aEW3k}
}
```

Privacy-preserving AUC computation in distributed machine learning with PHT-meDIC

PLOS Digital Health M Herr, CA Baykara, AB Ünal, N Pfeifer, M Akgün

Paper

BibTeX

@article{herr2025privacy,
  title   = {Privacy-preserving {AUC} computation in distributed machine learning with {PHT-meDIC}},
  author  = {de Arruda Botelho Herr, Marius and Baykara, Cem Ata and {\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  journal = {PLOS Digital Health},
  volume  = {4},
  number  = {11},
  pages   = {e0000753},
  year    = {2025},
  doi     = {10.1371/journal.pdig.0000753}
}

Accurate and Private Diagnosis of Rare Genetic Syndromes from Facial Images with Federated Deep Learning

arXiv preprint AB Ünal, CA Baykara, P Krawitz, M Akgün

Paper Code

BibTeX

@misc{unal2025accurate,
  title         = {Accurate and Private Diagnosis of Rare Genetic Syndromes from Facial Images with Federated Deep Learning},
  author        = {{\"U}nal, Ali Burak and Baykara, Cem Ata and Krawitz, Peter and Akg{\"u}n, Mete},
  year          = {2025},
  eprint        = {2509.10635},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2509.10635}
}

Federated Learning for Epileptic Seizure Prediction Across Heterogeneous EEG Datasets

arXiv preprint CA Baykara, SR Pandey, AB Ünal, H Lee, M Akgün

Paper

BibTeX

@misc{baykara2025federated,
  title         = {Federated Learning for Epileptic Seizure Prediction Across Heterogeneous {EEG} Datasets},
  author        = {Baykara, Cem Ata and Pandey, Saurav Raj and {\"U}nal, Ali Burak and Lee, Harlin and Akg{\"u}n, Mete},
  year          = {2025},
  eprint        = {2508.08159},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2508.08159}
}

2024

Dynamic k-anonymity for Electronic Health Records: A Topological Framework

DPM A Swaminathan, M Akgün

Paper Code

Abstract

With the rapid digitization of Electronic Health Records (EHRs), fast and adaptive data anonymization methods have become crucial. While topological data analysis (TDA) tools have been proposed to anonymize static datasets—creating multiple generalizations for different anonymization needs from a single computation—their application to dynamic datasets remains unexplored. Our work adapts existing methodologies to dynamic settings by developing an improved version of weighted persistence barcodes that track higher-dimensional holes in data, allowing real-time editing of persistence information.
BibTeX
```
@inproceedings{swaminathan2024dynamic,
  title     = {Dynamic k-Anonymity for Electronic Health Records: A Topological Framework},
  author    = {Swaminathan, Arjhun and Akg{\"u}n, Mete},
  booktitle = {Computer Security. ESORICS 2024 International Workshops (DPM)},
  series    = {Lecture Notes in Computer Science},
  volume    = {15263},
  pages     = {137--152},
  publisher = {Springer},
  year      = {2024},
  doi       = {10.1007/978-3-031-82349-7_10}
}
```
A privacy-preserving approach for cloud-based protein fold recognition

Patterns AB Ünal, N Pfeifer, M Akgün

Paper Code

Abstract

The complexity and cost of training machine learning models have made cloud-based machine learning as a service (MLaaS) attractive for businesses and researchers. MLaaS eliminates the need for in-house expertise by providing pre-built models and infrastructure. However, it raises data privacy and model security concerns, especially in medical fields like protein fold recognition. We propose a secure three-party computation-based MLaaS solution for privacy-preserving protein fold recognition, protecting both sequence and model privacy.
BibTeX
```
@article{unal2024privacy,
  title   = {A privacy-preserving approach for cloud-based protein fold recognition},
  author  = {{\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  journal = {Patterns},
  volume  = {5},
  number  = {9},
  pages   = {101023},
  year    = {2024},
  doi     = {10.1016/j.patter.2024.101023}
}
```
Private, Efficient and Scalable Kernel Learning for Medical Image Analysis

CIBB A Hannemann*, A Swaminathan*, AB Ünal, M Akgün

Paper

Abstract

Addressing the need for efficient privacy- preserving methods on distributed image data, we introduce OKRA (Or- thonormal K-fRAmes), a novel randomized encoding-based approach for kernel-based machine learning. This technique, tailored for widely used kernel functions, significantly enhances scalability and speed compared to current state-of-the-art solutions. Through experiments conducted on various clinical image datasets, we evaluated model quality, computa- tional performance, and resource overhead. Additionally, our method outperforms comparable approaches.
BibTeX
```
@inproceedings{hannemann2024private,
  title     = {Private, Efficient and Scalable Kernel Learning for Medical Image Analysis},
  author    = {Hannemann, Anika and Swaminathan, Arjhun and {\"U}nal, Ali Burak and Akg{\"u}n, Mete},
  booktitle = {Computational Intelligence Methods for Bioinformatics and Biostatistics (CIBB 2024)},
  series    = {Lecture Notes in Computer Science},
  volume    = {15276},
  publisher = {Springer},
  year      = {2024},
  doi       = {10.1007/978-3-031-89704-7_7}
}
```

Privacy Preserving Data Imputation via Multi-party Computation for Medical Applications

IEEE HealthCom J Jentsch, AB Ünal, ŞS Mağara, M Akgün

Paper

BibTeX

@inproceedings{jentsch2024privacy,
  title     = {Privacy Preserving Data Imputation via Multi-party Computation for Medical Applications},
  author    = {Jentsch, Julia and {\"U}nal, Ali Burak and Ma{\u{g}}ara, {\c{S}}eyma Selcan and Akg{\"u}n, Mete},
  booktitle = {2024 IEEE International Conference on E-health Networking, Application \& Services (HealthCom)},
  year      = {2024}
}

Leveraging Multi-Color Spaces as a Defense Mechanism Against Model Inversion Attack

ICML Workshop S Ouaari, AB Ünal, M Akgün, N Pfeifer

Paper

BibTeX

@inproceedings{ouaari2024leveraging,
  title     = {Leveraging Multi-Color Spaces as a Defense Mechanism Against Model Inversion Attack},
  author    = {Ouaari, Sofiane and {\"U}nal, Ali Burak and Akg{\"u}n, Mete and Pfeifer, Nico},
  booktitle = {ICML 2024 Workshop on the Next Generation of AI Safety},
  year      = {2024},
  url       = {https://icml.cc/virtual/2024/37498}
}

2023

A Privacy-Preserving Framework for Collaborative Machine Learning with Kernel Methods

IEEE TPS-ISA A Hannemann, AB Ünal, A Swaminathan, E Buchmann, M Akgün

Paper

Abstract

We’ve introduced FLAKE, a privacy-preserving Federated Learning Approach for Kernel methods on horizontally distributed data. By allowing data sources to mask their data, a centralized instance can generate a Gram matrix, preserving privacy while enabling the training of kernel-based algorithms like Support Vector Machines. We’ve established that FLAKE safeguards against semi-honest adversaries learning the input data or the number of features. Testing on clinical and synthetic data confirms FLAKE’s superior accuracy and efficiency over similar methods. Its data masking and Gram matrix computation times are significantly less than SVM training times, making it highly applicable across various use cases.
BibTeX
```
@inproceedings{hannemann2023privacy,
  title     = {A Privacy-Preserving Framework for Collaborative Machine Learning with Kernel Methods},
  author    = {Hannemann, Anika and {\"U}nal, Ali Burak and Swaminathan, Arjhun and Buchmann, Erik and Akg{\"u}n, Mete},
  booktitle = {2023 5th IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (TPS-ISA)},
  pages     = {82--90},
  year      = {2023},
  doi       = {10.1109/TPS-ISA58951.2023.00020}
}
```
ppAURORA: Privacy Preserving Area Under Receiver Operating Characteristic and Precision-Recall Curves

NSS AB Ünal, N Pfeifer, M Akgün

Paper Code

Abstract

In this paper, we propose an MPC-based framework, called PPAURORA with private merging of sorted lists and novel methods for comparing two secret-shared values, selecting between two secret-shared values, converting the modulus, and performing division to compute the exact AUC as one could obtain on the pooled original test samples.
BibTeX
```
@inproceedings{unal2023ppaurora,
  title     = {{ppAURORA}: Privacy Preserving Area Under Receiver Operating Characteristic and Precision-Recall Curves},
  author    = {{\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  booktitle = {Network and System Security (NSS 2023)},
  series    = {Lecture Notes in Computer Science},
  pages     = {265--280},
  publisher = {Springer},
  year      = {2023}
}
```

A Privacy-Preserving Scheme for Smart Grid Using Trusted Execution Environment

IEEE Access M Akgün, EU Soykan, G Soykan

Paper

BibTeX

@article{akgun2023privacy,
  title   = {A Privacy-Preserving Scheme for Smart Grid Using Trusted Execution Environment},
  author  = {Akg{\"u}n, Mete and Soykan, Elif Ustundag and Soykan, Gurkan},
  journal = {IEEE Access},
  volume  = {11},
  pages   = {9182--9196},
  year    = {2023},
  doi     = {10.1109/ACCESS.2023.3237643}
}

P/Key: PUF Based Second Factor Authentication

PLOS ONE E Uysal, M Akgün

Paper

BibTeX

@article{uysal2023pkey,
  title   = {{P/Key}: {PUF} based second factor authentication},
  author  = {Uysal, Ertan and Akg{\"u}n, Mete},
  journal = {PLOS ONE},
  volume  = {18},
  number  = {2},
  pages   = {e0280181},
  year    = {2023},
  doi     = {10.1371/journal.pone.0280181}
}

Scalable RFID Authentication Protocol Based on Physically Unclonable Functions

Computer Networks I Kurt, F Alagöz, M Akgün

Paper

BibTeX

@article{kurt2023scalable,
  title   = {Scalable {RFID} authentication protocol based on physically unclonable functions},
  author  = {Kurt, Isil and Alag{\"o}z, Fatih and Akg{\"u}n, Mete},
  journal = {Computer Networks},
  volume  = {230},
  pages   = {109793},
  year    = {2023},
  doi     = {10.1016/j.comnet.2023.109793}
}

2022

CECILIA: Comprehensive Secure Machine Learning Framework

arXiv preprint AB Ünal, N Pfeifer, M Akgün

Paper Code

Abstract

We propose a secure 3-party computation framework, CECILIA, offering PP building blocks to enable complex operations privately. In addition to the adapted and common operations like addition and multiplication, it offers multiplexer, most significant bit and modulus conversion. The first two are novel in terms of methodology and the last one is novel in terms of both functionality and methodology. CECILIA also has two complex novel methods, which are the exact exponential of a public base raised to the power of a secret value and the inverse square root of a secret Gram matrix.
BibTeX
```
@misc{unal2022cecilia,
  title         = {{CECILIA}: Comprehensive Secure Machine Learning Framework},
  author        = {{\"U}nal, Ali Burak and Pfeifer, Nico and Akg{\"u}n, Mete},
  year          = {2022},
  eprint        = {2202.03023},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2202.03023}
}
```
Efficient privacy-preserving whole-genome variant queries

Bioinformatics M Akgün, N Pfeifer, O Kohlbacher

Paper

Abstract

Disease–gene association studies are of great importance. However, genomic data are very sensitive when compared to other data types and contains information about individuals and their relatives. We propose a method that uses secure multi-party computation to query genomic databases in a privacy-protected manner.
BibTeX
```
@article{akgun2022efficient,
  title   = {Efficient privacy-preserving whole-genome variant queries},
  author  = {Akg{\"u}n, Mete and Pfeifer, Nico and Kohlbacher, Oliver},
  journal = {Bioinformatics},
  volume  = {38},
  number  = {8},
  pages   = {2202--2210},
  year    = {2022}
}
```

Bringing the Algorithms to the Data: Secure Distributed Medical Analytics using the Personal Health Train (PHT-meDIC)

arXiv preprint MAB Herr, M Graf, P Placzek, F König, F Bötte, T Stickel, D Hieber, L Zimmermann, M Slupina, C Mohr, S Biergans, M Akgün, N Pfeifer, O Kohlbacher

Paper

BibTeX

@misc{herr2022bringing,
  title         = {Bringing the Algorithms to the Data -- Secure Distributed Medical Analytics using the Personal Health Train ({PHT-meDIC})},
  author        = {de Arruda Botelho Herr, Marius and Graf, Michael and Placzek, Peter and K{\"o}nig, Florian and B{\"o}tte, Felix and Stickel, Tyra and Hieber, David and Zimmermann, Lukas and Slupina, Michael and Mohr, Christopher and Biergans, Stephanie and Akg{\"u}n, Mete and Pfeifer, Nico and Kohlbacher, Oliver},
  year          = {2022},
  eprint        = {2212.03481},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2212.03481}
}

2021

Escaped: Efficient secure and private dot product framework for kernel-based machine learning algorithms with applications in healthcare

AAAI AB Ünal, M Akgün, N Pfeifer

Paper Code

Abstract

We introduce ESCAPED, which stands for Efficient SeCure And PrivatE Dot product framework. ESCAPED enables the computation of the dot product of vectors from multiple sources on a third-party, which later trains kernel-based machine learning algorithms, while neither sacrificing privacy nor adding noise.
BibTeX
```
@inproceedings{unal2021escaped,
  title     = {{ESCAPED}: Efficient Secure and Private Dot Product Framework for Kernel-based Machine Learning Algorithms with Applications in Healthcare},
  author    = {{\"U}nal, Ali Burak and Akg{\"u}n, Mete and Pfeifer, Nico},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume    = {35},
  number    = {11},
  pages     = {9988--9996},
  year      = {2021},
  doi       = {10.1609/aaai.v35i11.17199}
}
```

2020

Identifying disease-causing mutations with privacy protection

Bioinformatics M Akgün, AB Ünal, B Ergüner, N Pfeifer, O Kohlbacher

Paper

Abstract

We present an approach to identify disease-associated variants and genes while ensuring patient privacy. The proposed method uses secure multi-party computation to find disease-causing mutations under specific inheritance models without sacrificing the privacy of individuals. It discloses only variants or genes obtained as a result of the analysis. Thus, the vast majority of patient data can be kept private.
BibTeX
```
@article{akgun2020identifying,
  title   = {Identifying disease-causing mutations with privacy protection},
  author  = {Akg{\"u}n, Mete and {\"U}nal, Ali Burak and Erg{\"u}ner, B. and Pfeifer, Nico and Kohlbacher, Oliver},
  journal = {Bioinformatics},
  volume  = {36},
  number  = {21},
  pages   = {5205--5213},
  year    = {2020},
  doi     = {10.1093/bioinformatics/btaa641}
}
```

Privacy-preserving SVM on Outsourced Genomic Data via Secure Multi-party Computation

IWSPA H Chen, AB Ünal, M Akgün, N Pfeifer

Paper

BibTeX

@inproceedings{chen2020privacy,
  title     = {Privacy-preserving {SVM} on Outsourced Genomic Data via Secure Multi-party Computation},
  author    = {Chen, H. and {\"U}nal, Ali Burak and Akg{\"u}n, Mete and Pfeifer, Nico},
  booktitle = {Proceedings of the Sixth International Workshop on Security and Privacy Analytics (IWSPA '20)},
  pages     = {61--69},
  publisher = {ACM},
  year      = {2020},
  doi       = {10.1145/3375708.3380316}
}

Privacy Preserving Gaze Estimation using Synthetic Images via a Randomized Encoding Based Framework

ACM ETRA E Bozkir, AB Ünal, M Akgün, E Kasneci, N Pfeifer

Paper

BibTeX

@inproceedings{bozkir2020privacy,
  title     = {Privacy Preserving Gaze Estimation using Synthetic Images via a Randomized Encoding Based Framework},
  author    = {Bozkir, Efe and {\"U}nal, Ali Burak and Akg{\"u}n, Mete and Kasneci, Enkelejda and Pfeifer, Nico},
  booktitle = {ACM Symposium on Eye Tracking Research and Applications (ETRA '20)},
  pages     = {1--5},
  publisher = {ACM},
  year      = {2020},
  doi       = {10.1145/3379156.3391364}
}

Eye Tracking Data Collection Protocol for VR for Remotely Located Subjects using Blockchain and Smart Contracts

IEEE AIVR E Bozkir, S Eivazi, M Akgün, E Kasneci

Paper

BibTeX

@inproceedings{bozkir2020eye,
  title     = {Eye Tracking Data Collection Protocol for {VR} for Remotely Located Subjects using Blockchain and Smart Contracts},
  author    = {Bozkir, Efe and Eivazi, Shahram and Akg{\"u}n, Mete and Kasneci, Enkelejda},
  booktitle = {2020 IEEE International Conference on Artificial Intelligence and Virtual Reality (AIVR)},
  year      = {2020},
  doi       = {10.1109/AIVR50618.2020.00083}
}

2019

A framework with randomized encoding for a fast privacy preserving calculation of non-linear kernels for machine learning applications in precision medicine

CANS AB Ünal, M Akgün, N Pfeifer

Paper

BibTeX

@inproceedings{unal2019framework,
  title     = {A framework with randomized encoding for a fast privacy preserving calculation of non-linear kernels for machine learning applications in precision medicine},
  author    = {{\"U}nal, Ali Burak and Akg{\"u}n, Mete and Pfeifer, Nico},
  booktitle = {Cryptology and Network Security (CANS 2019)},
  series    = {Lecture Notes in Computer Science},
  pages     = {493--511},
  publisher = {Springer},
  year      = {2019},
  doi       = {10.1007/978-3-030-31578-8_27}
}

An Active Genomic Data Recovery Attack

Balkan Journal of Electrical and Computer Engineering M Akgün

BibTeX

@article{akgun2019active,
  title   = {An Active Genomic Data Recovery Attack},
  author  = {Akg{\"u}n, Mete},
  journal = {Balkan Journal of Electrical and Computer Engineering},
  volume  = {7},
  number  = {4},
  pages   = {417--423},
  year    = {2019}
}