Sirvi Autor "Kamm, Liina, juhendaja" järgi

Nüüd näidatakse 1 - 4 4

An integrated approach for certification and re-certification based on the case study of an integrated circuit
(Tartu Ülikool, 2021) Thirumalai, Jayavarshini; Kamm, Liina, juhendaja; Seeba, Mari, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
A system is expected to undergo necessary security assessment to ensure that it is in compliance with the baseline security requirements. Otherwise it becomes hard to trust that the product is secure enough to use. For this purpose, certification can be used to ensure that a system is secure and safe to use. In this thesis, we define an integrated approach that aims to reduce time and cost in the product evaluation process by refining and integrating existing approaches. Hence, we consolidate solutions from the ARMOUR methodology, the ECSO meta-scheme and the NIST SP 800-137 to support certification and re-certification. We use a case study of the integrated circuit (or chip) as an example. In addition, we follow the Common Criteria based European Cybersecurity Candidate Scheme guidelines from ENISA to define a standardized process in certifying and re-certifying the product. Three different validators validated the thesis through face validity.
Eesti rahvastikuregistrile sarnase sünteetilise andmestiku reeglipõhine genereerimine
(Tartu Ülikool, 2025) Eichhorn, Rain; Laur, Sven, juhendaja; Kamm, Liina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The development and testing of information systems require realistic test data, especially in cases where privacy restrictions prohibit the use of real personal data. Rule-based synthetic data generation provides a way to create datasets that mimic the structure and logic of real data without processing personal information. In this thesis, I present the creation of a rule-based synthetic data generator for the Estonian Population Registry, relying solely on public input data and system business logic. The developed Python-based tool automatically generates use cases, enabling efficient and secure system testing. The results offer a practical foundation for future research and a privacy-preserving alternative to real data.
Framework for Privacy-Preserving Synthesis of Textual Data
(Tartu Ülikool, 2025) Stomakhin, Fedor; Laur, Sven, juhendaja; Kamm, Liina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
To safeguard patient privacy, sharing medical record data for research must adhere to various privacy regulations. To facilitate data sharing, various data protection techniques have been proposed, such as pseudonymization, anonymization and the use of synthetic data. The aim of synthetic data generation is, based on an original dataset, to produce a new dataset in a way that preserves the statistical relationships within the original data while not exposing any identifying or sensitive information about the data subjects therein. Synthetically generated data can still be insufficient from the point of view of privacy-preservation. For this purpose, approaches rooted in differential privacy (DP) have been proposed. DP typically relies on worst-case assumptions about attackers' knowledge, potentially leading to overly conservative measures. Applying DP principles to free-form text, such as medical epicrises, is complicated by their high dimensionality and complexity, as the same information can be conveyed in many different ways. In this work, motivated by the challenges of sharing textual health data, we propose and apply a general framework for evaluating privacy risks in text generated by large language models (LLMs). Considering a journalist attack model, we adapt differential privacy principles, quantifying privacy loss (ε, δ) based on the outputs of specific attack functions rather than relying on worst-case assumptions of DP. We demonstrate the framework by establishing baseline privacy characteristics via direct n-gram sampling analysis on both medical and social media texts and by exploring membership inference signals using surprisal analysis on LLMs fine-tuned with social media texts. While assessing synthetic data from standard LLMs highlighted methodological challenges, the framework provides a methodology for evaluating the privacy properties of text generation models and their outputs, informing decisions on sharing such data for research purposes.
Privacy-Preserving Data Synthesis Using Trusted Execution Environments
(Tartu Ülikool, 2022) Veskus, Karl Hannes; Kamm, Liina, juhendaja; Laur, Sven, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Data synthesis is the process of generating new synthetic data from existing data. Often companies do not have the the in-house competence to synthesize data themselves, and are willing to outsource the process. However, synthesis requires access to the original data. Sharing data with a third party can be complex, especially so if it contains sensitive information or is considered as personal data by regulations such as the GDPR. The goal of this thesis is to develop a proof-of-concept privacy-preserving data synthesis service showing that it is possible to use trusted execution environments to perform data synthesis in a privacy-preserving manner. Such a service would enable outsourcing the data synthesis process to an untrusted remote server by ensuring that both the original and synthesized data are fully hidden from the untrusted server host throughout the lifecycle of the service. A prototype of the service was developed in the scope of an ongoing proof-of-concept project. To achieve the required security goals the service prototype uses trusted execution environment technologies, specifically the Sharemind HI development platform, which is in turn based on the Intel SGX platform. The developed service shows that synthesizing data in a privacy-preserving manner is indeed feasible if trusted execution environments are used. However, future work is needed to optimize the service to allow larger input and output files, and to support additional data synthesis methods.