Visible to the public Biblio

Filters: Author is Kerschbaum, Florian  [Clear All Filters]
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 
Wang, Tianhao, Kerschbaum, Florian.  2019.  Attacks on Digital Watermarks for Deep Neural Networks. ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). :2622—2626.
Training deep neural networks is a computationally expensive task. Furthermore, models are often derived from proprietary datasets that have been carefully prepared and labelled. Hence, creators of deep learning models want to protect their models against intellectual property theft. However, this is not always possible, since the model may, e.g., be embedded in a mobile app for fast response times. As a countermeasure watermarks for deep neural networks have been developed that embed secret information into the model. This information can later be retrieved by the creator to prove ownership. Uchida et al. proposed the first such watermarking method. The advantage of their scheme is that it does not compromise the accuracy of the model prediction. However, in this paper we show that their technique modifies the statistical distribution of the model. Using this modification we can not only detect the presence of a watermark, but even derive its embedding length and use this information to remove the watermark by overwriting it. We show analytically that our detection algorithm follows consequentially from their embedding algorithm and propose a possible countermeasure. Our findings shall help to refine the definition of undetectability of watermarks for deep neural networks.
Fuhry, Benny, Tighzert, Walter, Kerschbaum, Florian.  2016.  Encrypting Analytical Web Applications. Proceedings of the 2016 ACM on Cloud Computing Security Workshop. :35–46.

The software-as-a-service (SaaS) market is growing very fast, but still many clients are concerned about the confidentiality of their data in the cloud. Motivated hackers or malicious insiders could try to steal the clients' data. Encryption is a potential solution, but supporting the necessary functionality also in existing applications is difficult. In this paper, we examine encrypting analytical web applications that perform extensive number processing operations in the database. Existing solutions for encrypting data in web applications poorly support such encryption. We employ a proxy that adjusts the encryption to the level necessary for the client's usage and also supports additively homomorphic encryption. This proxy is deployed at the client and all encryption keys are stored and managed there, while the application is running in the cloud. Our proxy is stateless and we only need to modify the database driver of the application. We evaluate an instantiation of our architecture on an exemplary application. We only slightly increase page load time on average from 3.1 seconds to 4.7. However, roughly 40% of all data columns remain probabilistic encrypted. The client can set the desired security level for each column using our policy mechanism. Hence our proxy architecture offers a solution to increase the confidentiality of the data at the cloud provider at a moderate performance penalty.

Hahn, Florian, Kerschbaum, Florian.  2016.  Poly-Logarithmic Range Queries on Encrypted Data with Small Leakage. Proceedings of the 2016 ACM on Cloud Computing Security Workshop. :23–34.

Privacy-preserving range queries allow encrypting data while still enabling queries on ciphertexts if their corresponding plaintexts fall within a requested range. This provides a data owner the possibility to outsource data collections to a cloud service provider without sacrificing privacy nor losing functionality of filtering this data. However, existing methods for range queries either leak additional information (like the ordering of the complete data set) or slow down the search process tremendously by requiring to query each ciphertext in the data collection. We present a novel scheme that only leaks the access pattern while supporting amortized poly-logarithmic search time. Our construction is based on the novel idea of enabling the cloud service provider to compare requested range queries. By doing so, the cloud service provider can use the access pattern to speed-up search time for range queries in the future. On the one hand, values that have fallen within a queried range, are stored in an interactively built index for future requests. On the other hand, values that have not been queried do not leak any information to the cloud service provider and stay perfectly secure. In order to show its practicability we have implemented our scheme and give a detailed runtime evaluation.

Hahn, Florian, Loza, Nicolas, Kerschbaum, Florian.  2018.  Practical and Secure Substring Search. Proceedings of the 2018 International Conference on Management of Data. :163–176.
In this paper we address the problem of outsourcing sensitive strings while still providing the functionality of substring searches. While security is one important aspect that requires careful system design, the practical application of the solution depends on feasible processing time and integration efforts into existing systems. That is, searchable symmetric encryption (SSE) allows queries on encrypted data but makes common indexing techniques used in database management systems for fast query processing impossible. As a result, the overhead for deploying such functional and secure encryption schemes into database systems while maintaining acceptable processing time requires carefully designed special purpose index structures. Such structures are not available on common database systems but require individual modifications depending on the deployed SSE scheme. Our technique transforms the problem of secure substring search into range queries that can be answered efficiently and in a privacy-preserving way on common database systems without further modifications using frequency-hiding order-preserving encryption. We evaluated our prototype implementation deployed in a real-world scenario, including the consideration of network latency, we demonstrate the practicability of our scheme with 98.3 ms search time for 10,000 indexed emails. Further, we provide a practical security evaluation of this transformation based on the bucketing attack that is the best known published attack against this kind of property-preserving encryption.