A5090704762- Cryptography and Data Security
- Privacy-Preserving Technologies in Data
- Speech Recognition and Synthesis
- Internet Traffic Analysis and Secure E-voting
- Blockchain Technology Applications and Security
- Music and Audio Processing
- Speech and Audio Processing
- Adversarial Robustness in Machine Learning
- Cryptographic Implementations and Security
- Stochastic Gradient Optimization Techniques
- Chaos-based Image/Signal Encryption
- User Authentication and Security Systems
- Advanced Malware Detection Techniques
- Software System Performance and Reliability
- Cloud Data Security Solutions
- Biometric Identification and Security
- Imbalanced Data Classification Techniques
- Network Security and Intrusion Detection
- Advanced Data Storage Technologies
- Authorship Attribution and Profiling
- Distributed systems and fault tolerance
Technical University of Darmstadt
2018-2024
Brown University
2023
Université Mohammed VI Polytechnique
2023
Polytechnic University
2023
Carnegie Mellon University
2023
Speech recordings are a rich source of personal, sensitive data that can be used to support plethora diverse applications, from health profiling biometric recognition. It is therefore essential speech adequately protected so they cannot misused. Such protection, in the form privacy-preserving technologies, required ensure that: (i) profiles given individual (e.g., across different service operators) unlinkable; (ii) leaked, encrypted information irreversible, and (iii) references renewable....
An encrypted search algorithm (ESA) allows a user to encrypt its data while preserving the ability over it. As all practical solutions leak some information, cryptanalysis plays an important role in area of search. Starting with work Islam et al. (NDSS'12), many attacks have been proposed that exploit different leakage profiles under various assumptions. While these improve our understanding leakage, it can sometimes be difficult draw definite conclusions about their performance. This is due...
Encrypted search algorithms (ESAs) enable private on encrypted data and can be constructed from a variety of cryptographic primitives. All knownsub-linear ESA leak information and, therefore, the design leakage attacks is an important way to ascertain whether given profile exploitable in practice. Recently,Oya Kerschbaum(Usenix '22) presented attack called IHOP that targets query equality pattern which reveals if when two queries are for same keyword sequence dependent queries. In this work,...
Pushes for increased power of Law Enforcement (LE) data retention and centralized storage result in legal challenges with protection law courts-and possible violations the right to privacy. This is motivated by a desire better cooperation exchange between LE Agencies (LEAs), which difficult due regulations, was identified as main factor major public security failures, frequent criticism LE. Secure Multi-Party Computation (MPC) often seen technological means solve privacy conflicts where...
In many voice biometrics applications there is a requirement to preserve privacy, not least because of the recently enforced General Data Protection Regulation (GDPR).Though progress in bringing privacy preservation lagging behind developments other communities, recent years have seen rapid progress, with secure computation mechanisms such as homomorphic encryption being applied successfully speaker recognition.Even so, computational overhead incurred by processing speech data encrypted...
In this work, we provide an industry research view for approaching the design, deployment, and operation of trustworthy Artificial Intelligence (AI) inference systems. Such systems customers with timely, informed, customized inferences to aid their decision, while at same time utilizing appropriate security protection mechanisms AI models. Additionally, such should also use Privacy-Enhancing Technologies (PETs) protect customers' data any time. To approach subject, start by introducing...
AI algorithms, and machine learning (ML) techniques in particular, are increasingly important to individuals' lives, but have caused a range of privacy concerns addressed by, e.g., the European GDPR. Using cryptographic techniques, it is possible perform inference tasks remotely on sensitive client data privacy-preserving way: server learns nothing about input model predictions, while ML (which often considered intellectual property might contain traces data). While such solutions relatively...
The ubiquitous deployment of machine learning (ML) technologies has certainly improved many applications but also raised challenging privacy concerns, as sensitive client data is usually processed remotely at the discretion a service provider. Therefore, privacy-preserving (PPML) aims providing using techniques such secure multi-party computation (SMPC).
Encrypted Search Algorithms (ESAs) are a technique to encrypt data while the user can still search over it. ESAs protect privacy and ensure security of sensitive stored on remote storage. Originally, were used in context documents that consist keywords. The encrypts documents, sends them server is able for keywords, without exposing information about plaintext. idea has also been applied relational databases, where queries (similar SQL statements) be privately executed an encrypted database....
Privacy-preserving scalar product (PPSP) protocols are an important building block for secure computation tasks in various applications. Lu et al. (TPDS'13) introduced a PPSP protocol that does not rely on cryptographic assumptions and is used wide range of publications to date. In this comment paper, we show al.'s insecure should be used. We describe specific attacks against it and, using impossibility results Impagliazzo Rudich (STOC'89), inherently cannot fixed without relying at least...
The well-defined information leakage of Encrypted Search Algorithms (ESAs) is predominantly analyzed by crafting so-called attacks. These attacks utilize adversarially known auxiliary data and the observed to attack an ESA instance built on a user's data. Known-data require be subset In contrast, sampled-data merely rely that is, in some sense, statistically close hence reflect much more realistic scenario where stems from publicly available source instead private
Oblivious RAM is a cryptographic primitive that embodies one of the cornerstones privacy-preserving technologies for database protection. While any (ORAM) construction offers access pattern hiding, there does not seem to be safe against potential leakage due knowledge about number accesses performed by client. Such constitutes privacy violation, as client data may stored in domain specific fashion. In this work, we examine considering an adversary can probe server stores ORAM database, and...
In many voice biometrics applications there is a requirement to preserve privacy, not least because of the recently enforced General Data Protection Regulation (GDPR). Though progress in bringing privacy preservation lagging behind developments other communities, recent years have seen rapid progress, with secure computation mechanisms such as homomorphic encryption being applied successfully speaker recognition. Even so, computational overhead incurred by processing speech data encrypted...