Helmholtz Center for Information Security
CISPA – Helmholtz-Zentrum für InformationssicherheitNot a member yet
3406 research outputs found
Sort by
Sequential Attacks on Agents for Long-Term Adversarial Goals
Get PDFReinforcement learning (RL) has advanced greatly in the past few years with the employment of effective deep neural networks (DNNs) on the policy networks. With the great effectiveness came serious vulnerability issues with DNNs that small adversarial perturbations on the input can change the output of the network. Several works have pointed out that learned agents with a DNN policy network can be manipulated against achieving the original task through a sequence of small perturbations on the input states. In this paper, we demonstrate furthermore that it is also possible to impose an arbitrary adversarial reward on the victim policy network through a sequence of attacks. Our method involves the latest adversarial attack technique, Adversarial Transformer Network (ATN), that learns to generate the attack and is easy to integrate into the policy network. As a result of our attack, the victim agent is misguided to optimise for the adversarial reward over time. Our results expose serious security threats for RL applications in safety-critical systems including drones, medical analysis, and self-driving cars
Paris Call: Important Commitment Against The Arms Race With Cyber Weapons
Get PDFThe following text by Ninja Marnau was published on 14.12.2018 in the Tagesspiegel - Background Digitalisierung (in a shortened version) under the title "Disclosing security vulnerabilities instead of selling them"
Rule Discovery for Exploratory Causal Reasoning
Get PDFWe study the problem of discovering reliable causal rules from observational data.
Traditional descriptive rule discovery techniques do not suffice to this end, as
they struggle with the consistent detection of (potentially rare) conditions that
have a strong effect on an output variable of interest. Among the sources of
inconsistency are that naive empirical effect estimations have a high variance,
and, hence, their maximization is highly optimistically biased unless the search
is artificially restricted to high frequency events. Secondly, observational effect
measurements are often highly unrepresentative of the underlying causal effect
because they are skewed by the presence of confounding factors. This is a concern
especially in scientific data analysis.
To address these issues, we present a novel descriptive rule discovery approach
based on reliably estimating the conditional effect given the potential confounders.
We demonstrate that the corresponding score is a conservative and consistent effect
estimator, identify the admissible data generation process under which causal
rule discovery is possible, and derive an efficient optimization algorithm that
successfully detects valuable rules on a multitude of real datasets. Important for
both causal and associational data exploration, the presented approach naturally
allows for iterative rule discovery, where new non-redundant rules can be found by
treating previously discovered rules as confounders in subsequent iterations
DRiVERSITY - Synthetic Torture Testing to Find Limits of Autonomous Driving Algorithms
No full textAutonomous driving is expected to significantly improve road safety. Carmakers are conducting extensive testing of their autonomous vehicles on proofing grounds and in virtual pre-defined scenarios. Because proofing grounds do not offer a deterministic test field and are time-consuming, virtual hardware- and software-in-the-loop testing is en vogue as it provides reproducibility. However, pre-defined tests (real or virtual) represent low coverage in comparison to all physically possible driving scenarios. Furthermore, they are unlikely to systematically discover corner cases that emerge due to software bugs or absurd but possible scenarios.
In this abstract, we introduce the DRiVERSITY framework for systematic testing of autonomous driving algorithms. Our DRiVERSITY framework builds scenes on the fly – adapting to how a car handles situations while driving. DRiVERSITY pronounces misbehaviour by tailoring new scenes based on monitored driving behaviour during fuzzing stimuli. DRiVERSITY shall provide a standard testing framework to evaluate and compare driving algorithms in a reproducible and controllable way
Analyzing Guarded Protocols: Better Cutoffs, More Systems, More Expressivity
Get PDFWe study cutoff results for parameterized verification and synthesis of guarded protocols, as introduced by Emerson and Kahlon (2000). Guarded protocols describe systems of processes whose transitions are enabled or disabled depending on the existence of other processes in certain local states. Cutoff results reduce reasoning about systems with an arbitrary number of processes to systems of a determined, fixed size. Our work is based on the observation that existing cutoff results for guarded protocols are often impractical, since they scale linearly in the number of local states of processes in the system. We provide new cutoffs that scale not with the number of local states, but with the number of guards in the system, which is in many cases much smaller. Furthermore, we consider generalizations of the type of guards and of the specifications under consideration, and present results for problems that have not been known to admit cutoffs before
Formally Reasoning about the Cost and Efficacy of Securing the Email Infrastructure
Get PDFSecurity in the Internet has historically been added post-hoc, leaving services like email, which, after all, is used by 3.7 billion users, vulnerable to large-scale surveillance. For email alone, there is a multitude of proposals to mitigate known vulnerabilities, ranging from the introduction of completely new protocols to modifications of the communication paths used by big providers. Deciding which measures to deploy requires a deep understanding of the induced benefits, the cost and the resulting effects. This paper proposes the first automated methodology for making formal deployment assessments. Our planning algorithm analyses the impact and cost-efficiency of different known mitigation strategies against an attacker in a formal threat model. This novel formalisation of an infrastructure attacker includes routing, name resolution and application level weaknesses. We apply the methodology to a large-scale scan of the Internet, and assess how protocols like IPsec, DNSSEC, DANE, SMTP STS, SMTP over TLS and other mitigation techniques like server relocation can be combined to improve the confidentiality of email users in 45 combinations of attacker and defender countries and nine cost scenarios. This is the first deployment analysis for mitigation techniques at this scale
teEther: Gnawing at Ethereum to Automatically Exploit Smart Contracts
Get PDFCryptocurrencies like Bitcoin not only provide a decentralized currency, but also provide a programmatic way to process transactions. Ethereum, the second largest cryptocurrency next to Bitcoin, is the first to provide a Turing-complete language to specify transaction processing, thereby enabling so-called smart contracts. This provides an opportune setting for attackers, as security vulnerabilities are tightly intertwined with financial gain. In this paper, we consider the problem of automatic vulnerability identification and exploit generation for smart contracts. We develop a generic definition of vulnerable contracts and use this to build TEE THER, a tool that allows creating an exploit for a contract given only its binary bytecode. We perform a large-scale analysis of all 38,757 unique Ethereum contracts, 815 out of which our tool finds working exploits for—completely automated