FPFelix Pauck

Felix Pauck

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Felix Pauck has a total of 6 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Scaling Arbitrary Android App Analyses

    by Felix Pauck

    More apps are published every day and the functionality of each app increases steadily as well. Consequently app analyses are often overwhelmed when confronted with up-to-date, real-world apps. One of the biggest issues originates from the scalability of analyses with respect to libraries. Analyses, more precisely the tools implementing them, cannot distinguish the app’s code from the code of a library. Always analyzing the whole code base is the result. However, this is usually not necessary, for example, when a security property is checked, trusted libraries must not be analyzed. We propose an approach to differentiate an app’s code from a library’s code. The approach is based on clone detection and implemented in our prototype APK-Simplifier. As the evaluation shows APK-Simplifier can be employed in a cooperative analysis to remove library code and to enhance arbitrary analysis tools’ scalability. In fact, five analysis tools have been enabled to analyze five up-to-date, real-world apps they could not analyze before. Still, it is alerting that the majority of such apps remains not analyzable as also shown during evaluation.


  • Open Access

    Are Neural Bug Detectors Comparable to Software Developers on Variable Misuse Bugs?

    by Cedric Richter, Jan Haltermann, Marie-Christine Jakobs, Felix Pauck, Stefan Schott, Heike Wehrheim

    Debugging, that is, identifying and fixing bugs in software, is a central part of software development. Developers are therefore often confronted with the task of deciding whether a given code snippet contains a bug, and if yes, where. Recently, data-driven methods have been employed to learn this task of bug detection, resulting (amongst others) in so called neural bug detectors. Neural bug detectors are trained on millions of buggy and correct code snippets. Given the “neural learning” procedure, it seems likely that neural bug detectors – on the specific task of finding bugs – have a performance similar to human software developers. For this work, we set out to substantiate or refute such a hypothesis. We report on the results of an empirical study with over 100 software developers, targeting the comparison of humans and neural bug detectors. As detection task, we chose a specific form of bugs (variable misuse bugs) for which neural bug detectors have recently made significant progress. Our study shows that despite the fact that neural bug detectors see millions of such misuse bugs during training, software developers – when conducting bug detection as a majority decision – are slightly better than neural bug detectors on this class of bugs. Altogether, we find a large overlap in the performance, both for classifying code as buggy and for localizing the buggy line in the code. In comparison to developers, one of the two evaluated neural bug detectors, however, raises a higher number of false alarms in our study.


  • Android App Merging for Benchmark Speed-Up and Analysis Lift-Up

    by Felix Pauck, Shikun Zhang

    In the field of software analysis a trade-off between scalability and accuracy always exists. In this respect, Android app analysis is no exception, in particular, analyzing large or many apps can be challenging. Dealing with many small apps is a typical challenge when facing micro-benchmarks such as DROIDBENCH or ICC-BENCH. These particular benchmarks are not only used for the evaluation of novel tools but also in continuous integration pipelines of existing mature tools to maintain and guarantee a certain quality-level. Considering this latter usage it becomes very important to be able to achieve benchmark results as fast as possible. Hence, benchmarks have to be optimized for this purpose. One approach to do so is app merging. We implemented the Android Merge Tool (AMT) following this approach and show that its novel aspects can be used to produce scaled up and accurate benchmarks. For such benchmarks Android app analysis tools do not suffer from the scalability-accuracy trade-off anymore. We show this throughout detailed experiments on DROIDBENCH employing three different analysis tools (AMANDROID, ICCTA, FLOWDROID). Benchmark execution times are largely reduced without losing benchmark accuracy. Moreover, we argue why AMT is an advantageous successor of the state-of-the-art app merging tool (APKCOMBINER) in analysis lift-up scenarios.