Helmholtz Center for Information Security
CISPA – Helmholtz-Zentrum für InformationssicherheitNot a member yet
3406 research outputs found
Sort by
A monadic framework for relational verification: applied to information security, program equivalence, and optimizations
No full textAdvanced Steel Microstructural Classification by Deep Learning Methods
No full textThe inner structure of a material is called microstructure. It stores the genesis of a material and determines all its physical and chemical properties. While microstructural characterization is widely spread and well known, the microstructural classification is mostly done manually by human experts, which gives rise to uncertainties due to subjectivity. Since the microstructure could be a combination of different phases or constituents with complex substructures its automatic classification is very challenging and only a few prior studies exist. Prior works focused on designed and engineered features by experts and classified microstructures separately from the feature extraction step. Recently, Deep Learning methods have shown strong performance in vision applications by learning the features from data together with the classification step. In this work, we propose a Deep Learning method for microstructural classification in the examples of certain microstructural constituents of low carbon steel. This novel method employs pixel-wise segmentation via Fully Convolutional Neural Network (FCNN) accompanied by a max-voting scheme. Our system achieves 93.94% classification accuracy, drastically outperforming the state-of-the-art method of 48.89% accuracy. Beyond the strong performance of our method, this line of research offers a more robust and first of all objective way for the difficult task of steel quality appreciation
Transferring Tests Across Web Applications
No full textWhen manually testing Web applications, humans can go with vague, yet general instructions, such as ``add the product to shopping cart and proceed to checkout''. Can we teach a robot to follow such instructions as well? We show how to leverage tests of other applications to guide test generation for new applications in the same domain: Given a test for payments on Amazon, we guide test generation on eBay towards payment functionality, exploiting the semantic similarity between UI elements across both applications. Evaluated on twelve Web apps in three domains, our approach allows for discovering deep functionality in a few minutes, where an undirected crawler would require days or weeks to accomplish the same task
A4NT: Author Attribute Anonymity by Adversarial Training of Neural Machine Translation
No full textText-based analysis methods enable an adversary to reveal privacy relevant author attributes such as gender, age and can identify the text's author. Such methods can compromise the privacy of an anonymous author even when the author tries to remove privacy sensitive content. In this paper, we propose an automatic method, called the Adversarial Author Attribute Anonymity Neural Translation (), to combat such text-based adversaries. Unlike prior works on obfuscation, we propose a system that is fully automatic and learns to perform obfuscation entirely from the data. This allows us to easily apply the system to obfuscate different author attributes. We propose a sequence-to-sequence language model, inspired by machine translation, and an adversarial training framework to design a system which learns to transform the input text to obfuscate the author attributes without paired data. We also propose and evaluate techniques to impose constraints on our model to preserve the semantics of the input text. learns to make minimal changes to the input to successfully fool author attribute classifiers, while preserving the meaning of the input text. Our experiments on two datasets and three settings show that the proposed method is effective in fooling the attribute classifiers and thus improves the anonymity of authors
Wie sicher ist das beA? – Das besondere elektronische Anwaltspostfach aus Sicht der IT-Sicherheit
No full textAnswering Visual What-If Questions: From Actions to Predicted Scene Descriptions
No full textIn-depth scene descriptions and question answering tasks have greatly increased the scope of today's definition of scene understanding. While such tasks are in principle open ended, current formulations primarily focus on describing only the current state of the scenes under consideration. In contrast, in this paper, we focus on the future states of the scenes which are also conditioned on actions. We posit this as a question answering task, where an answer has to be given about a future scene state, given observations of the current scene, and a question that includes a hypothetical action. Our solution is a hybrid model which integrates a physics engine into a question answering architecture in order to anticipate future scene states resulting from object-object interactions caused by an action. We demonstrate first results on this challenging new problem and compare to baselines, where we outperform fully data-driven end-to-end learning approaches
JAMI: Fast Computation of Conditional Mutual Information for ceRNA network analysis
Get PDFMotivation: Genome-wide measurements of paired miRNA and gene expression data have
enabled the prediction of competing endogenous RNAs (ceRNAs). It has been shown that the
sponge effect mediated by protein-coding as well as non-coding ceRNAs can play an important
regulatory role in the cell in health and disease. Therefore, many computational methods for the
computational identification of ceRNAs have been suggested. In particular, methods based on
Conditional Mutual Information (CMI) have shown promising results. However, the currently
available implementation is slow and cannot be used to perform computations on a large scale.
Results: Here, we present JAMI, a Java tool that uses a non-parametric estimator for CMI
values from gene and miRNA expression data. We show that JAMI speeds up the computation of
ceRNA networks by a factor of 70 compared to currently available implementations. Further,
JAMI supports multi-threading to make use of common multi-core architectures for further
performance gain
Stochastic Complexity for Testing Conditional Independence on Discrete Data
Get PDFTesting for conditional independence is a core aspect of constraint-based causal discovery. Although commonly used tests are perfect in theory, they often fail to reject independence in practice, especially when conditioning on multiple variables.
We focus on discrete data and propose a new test based on the notion of algorithmic independence that we instantiate using stochastic complexity. Amongst others, we show that our proposed test, SCI, is an asymptotically unbiased as well as L2 consistent estimator for conditional mutual information (CMI). Further, we show that SCI can be reformulated to find a sensible threshold for CMI that works well on limited samples. Empirical evaluation shows that SCI has a lower type II error than commonly used tests. As a result, we obtain a higher recall when we use SCI in causal discovery algorithms, without compromising the precision
Evasive Malware via Identifier Implanting
Get PDFTo cope with the increasing number of malware attacks that organizations face, anti-malware appliances and sandboxes have become an integral security defense. In particular, appliances have become the de facto standard in the fight against targeted attacks. Yet recent incidents have demonstrated that malware can effectively detect and thus evade sandboxes, resulting in an ongoing arms race between sandbox developers and malware authors.
We show how attackers can escape this arms race with what we call customized malware, i.e., malware that only exposes its malicious behavior on a targeted system. We present a web-based reconnaissance strategy, where an actor leaves marks on the target system such that the customized malware can recognize this particular system in a later stage, and only then exposes its malicious behavior. We propose to implant identifiers into the target system, such as unique entries in the browser history, cache, cookies, or the DNS stub resolver cache. We then prototype a customized malware that searches for these implants on the executing environment and denies execution if implants do not exist as expected. This way, sandboxes can be evaded without the need to detect artifacts that witness the existence of sandboxes or a real system environment. Our results show that this prototype remains undetected on commercial malware security appliances, while only exposing its real behavior on the targeted system. To defend against this novel attack, we discuss countermeasures and a responsible disclosure process to allow appliances vendors to prepare for such attacks