@inproceedings{12854, abstract = {The main idea behind BUBAAK is to run multiple program analyses in parallel and use runtime monitoring and enforcement to observe and control their progress in real time. The analyses send information about (un)explored states of the program and discovered invariants to a monitor. The monitor processes the received data and can force an analysis to stop the search of certain program parts (which have already been analyzed by other analyses), or to make it utilize a program invariant found by another analysis. At SV-COMP 2023, the implementation of data exchange between the monitor and the analyses was not yet completed, which is why BUBAAK only ran several analyses in parallel, without any coordination. Still, BUBAAK won the meta-category FalsificationOverall and placed very well in several other (sub)-categories of the competition.}, author = {Chalupa, Marek and Henzinger, Thomas A}, booktitle = {Tools and Algorithms for the Construction and Analysis of Systems}, isbn = {9783031308192}, issn = {1611-3349}, location = {Paris, France}, pages = {535--540}, publisher = {Springer Nature}, title = {{Bubaak: Runtime monitoring of program verifiers}}, doi = {10.1007/978-3-031-30820-8_32}, volume = {13994}, year = {2023}, } @inproceedings{12856, abstract = {As the complexity and criticality of software increase every year, so does the importance of run-time monitoring. Third-party monitoring, with limited knowledge of the monitored software, and best-effort monitoring, which keeps pace with the monitored software, are especially valuable, yet underexplored areas of run-time monitoring. Most existing monitoring frameworks do not support their combination because they either require access to the monitored code for instrumentation purposes or the processing of all observed events, or both. We present a middleware framework, VAMOS, for the run-time monitoring of software which is explicitly designed to support third-party and best-effort scenarios. The design goals of VAMOS are (i) efficiency (keeping pace at low overhead), (ii) flexibility (the ability to monitor black-box code through a variety of different event channels, and the connectability to monitors written in different specification languages), and (iii) ease-of-use. To achieve its goals, VAMOS combines aspects of event broker and event recognition systems with aspects of stream processing systems. We implemented a prototype toolchain for VAMOS and conducted experiments including a case study of monitoring for data races. The results indicate that VAMOS enables writing useful yet efficient monitors, is compatible with a variety of event sources and monitor specifications, and simplifies key aspects of setting up a monitoring system from scratch.}, author = {Chalupa, Marek and Mühlböck, Fabian and Muroya Lei, Stefanie and Henzinger, Thomas A}, booktitle = {Fundamental Approaches to Software Engineering}, isbn = {9783031308253}, issn = {1611-3349}, location = {Paris, France}, pages = {260--281}, publisher = {Springer Nature}, title = {{Vamos: Middleware for best-effort third-party monitoring}}, doi = {10.1007/978-3-031-30826-0_15}, volume = {13991}, year = {2023}, } @misc{12407, abstract = {As the complexity and criticality of software increase every year, so does the importance of run-time monitoring. Third-party monitoring, with limited knowledge of the monitored software, and best-effort monitoring, which keeps pace with the monitored software, are especially valuable, yet underexplored areas of run-time monitoring. Most existing monitoring frameworks do not support their combination because they either require access to the monitored code for instrumentation purposes or the processing of all observed events, or both. We present a middleware framework, VAMOS, for the run-time monitoring of software which is explicitly designed to support third-party and best-effort scenarios. The design goals of VAMOS are (i) efficiency (keeping pace at low overhead), (ii) flexibility (the ability to monitor black-box code through a variety of different event channels, and the connectability to monitors written in different specification languages), and (iii) ease-of-use. To achieve its goals, VAMOS combines aspects of event broker and event recognition systems with aspects of stream processing systems. We implemented a prototype toolchain for VAMOS and conducted experiments including a case study of monitoring for data races. The results indicate that VAMOS enables writing useful yet efficient monitors, is compatible with a variety of event sources and monitor specifications, and simplifies key aspects of setting up a monitoring system from scratch.}, author = {Chalupa, Marek and Mühlböck, Fabian and Muroya Lei, Stefanie and Henzinger, Thomas A}, issn = {2664-1690}, keywords = {runtime monitoring, best effort, third party}, pages = {38}, publisher = {Institute of Science and Technology Austria}, title = {{VAMOS: Middleware for Best-Effort Third-Party Monitoring}}, doi = {10.15479/AT:ISTA:12407}, year = {2023}, } @inproceedings{13142, abstract = {Reinforcement learning has received much attention for learning controllers of deterministic systems. We consider a learner-verifier framework for stochastic control systems and survey recent methods that formally guarantee a conjunction of reachability and safety properties. Given a property and a lower bound on the probability of the property being satisfied, our framework jointly learns a control policy and a formal certificate to ensure the satisfaction of the property with a desired probability threshold. Both the control policy and the formal certificate are continuous functions from states to reals, which are learned as parameterized neural networks. While in the deterministic case, the certificates are invariant and barrier functions for safety, or Lyapunov and ranking functions for liveness, in the stochastic case the certificates are supermartingales. For certificate verification, we use interval arithmetic abstract interpretation to bound the expected values of neural network functions.}, author = {Chatterjee, Krishnendu and Henzinger, Thomas A and Lechner, Mathias and Zikelic, Dorde}, booktitle = {Tools and Algorithms for the Construction and Analysis of Systems }, isbn = {9783031308222}, issn = {1611-3349}, location = {Paris, France}, pages = {3--25}, publisher = {Springer Nature}, title = {{A learner-verifier framework for neural network controllers and certificates of stochastic systems}}, doi = {10.1007/978-3-031-30823-9_1}, volume = {13993}, year = {2023}, } @inproceedings{12467, abstract = {Safety and liveness are elementary concepts of computation, and the foundation of many verification paradigms. The safety-liveness classification of boolean properties characterizes whether a given property can be falsified by observing a finite prefix of an infinite computation trace (always for safety, never for liveness). In quantitative specification and verification, properties assign not truth values, but quantitative values to infinite traces (e.g., a cost, or the distance to a boolean property). We introduce quantitative safety and liveness, and we prove that our definitions induce conservative quantitative generalizations of both (1)~the safety-progress hierarchy of boolean properties and (2)~the safety-liveness decomposition of boolean properties. In particular, we show that every quantitative property can be written as the pointwise minimum of a quantitative safety property and a quantitative liveness property. Consequently, like boolean properties, also quantitative properties can be min-decomposed into safety and liveness parts, or alternatively, max-decomposed into co-safety and co-liveness parts. Moreover, quantitative properties can be approximated naturally. We prove that every quantitative property that has both safe and co-safe approximations can be monitored arbitrarily precisely by a monitor that uses only a finite number of states.}, author = {Henzinger, Thomas A and Mazzocchi, Nicolas Adrien and Sarac, Naci E}, booktitle = {26th International Conference Foundations of Software Science and Computation Structures}, isbn = {9783031308284}, issn = {1611-3349}, location = {Paris, France}, pages = {349--370}, publisher = {Springer Nature}, title = {{Quantitative safety and liveness}}, doi = {10.1007/978-3-031-30829-1_17}, volume = {13992}, year = {2023}, } @inproceedings{13292, abstract = {The operator precedence languages (OPLs) represent the largest known subclass of the context-free languages which enjoys all desirable closure and decidability properties. This includes the decidability of language inclusion, which is the ultimate verification problem. Operator precedence grammars, automata, and logics have been investigated and used, for example, to verify programs with arithmetic expressions and exceptions (both of which are deterministic pushdown but lie outside the scope of the visibly pushdown languages). In this paper, we complete the picture and give, for the first time, an algebraic characterization of the class of OPLs in the form of a syntactic congruence that has finitely many equivalence classes exactly for the operator precedence languages. This is a generalization of the celebrated Myhill-Nerode theorem for the regular languages to OPLs. As one of the consequences, we show that universality and language inclusion for nondeterministic operator precedence automata can be solved by an antichain algorithm. Antichain algorithms avoid determinization and complementation through an explicit subset construction, by leveraging a quasi-order on words, which allows the pruning of the search space for counterexample words without sacrificing completeness. Antichain algorithms can be implemented symbolically, and these implementations are today the best-performing algorithms in practice for the inclusion of finite automata. We give a generic construction of the quasi-order needed for antichain algorithms from a finite syntactic congruence. This yields the first antichain algorithm for OPLs, an algorithm that solves the ExpTime-hard language inclusion problem for OPLs in exponential time.}, author = {Henzinger, Thomas A and Kebis, Pavol and Mazzocchi, Nicolas Adrien and Sarac, Naci E}, booktitle = {50th International Colloquium on Automata, Languages, and Programming}, isbn = {9783959772785}, issn = {1868-8969}, location = {Paderborn, Germany}, pages = {129:1----129:20}, publisher = {Schloss Dagstuhl - Leibniz-Zentrum für Informatik}, title = {{Regular methods for operator precedence languages}}, doi = {10.4230/LIPIcs.ICALP.2023.129}, volume = {261}, year = {2023}, } @article{12704, abstract = {Adversarial training (i.e., training on adversarially perturbed input data) is a well-studied method for making neural networks robust to potential adversarial attacks during inference. However, the improved robustness does not come for free but rather is accompanied by a decrease in overall model accuracy and performance. Recent work has shown that, in practical robot learning applications, the effects of adversarial training do not pose a fair trade-off but inflict a net loss when measured in holistic robot performance. This work revisits the robustness-accuracy trade-off in robot learning by systematically analyzing if recent advances in robust training methods and theory in conjunction with adversarial robot learning, are capable of making adversarial training suitable for real-world robot applications. We evaluate three different robot learning tasks ranging from autonomous driving in a high-fidelity environment amenable to sim-to-real deployment to mobile robot navigation and gesture recognition. Our results demonstrate that, while these techniques make incremental improvements on the trade-off on a relative scale, the negative impact on the nominal accuracy caused by adversarial training still outweighs the improved robustness by an order of magnitude. We conclude that although progress is happening, further advances in robust learning methods are necessary before they can benefit robot learning tasks in practice.}, author = {Lechner, Mathias and Amini, Alexander and Rus, Daniela and Henzinger, Thomas A}, issn = {2377-3766}, journal = {IEEE Robotics and Automation Letters}, number = {3}, pages = {1595--1602}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{Revisiting the adversarial robustness-accuracy tradeoff in robot learning}}, doi = {10.1109/LRA.2023.3240930}, volume = {8}, year = {2023}, } @inproceedings{14242, abstract = {We study the problem of training and certifying adversarially robust quantized neural networks (QNNs). Quantization is a technique for making neural networks more efficient by running them using low-bit integer arithmetic and is therefore commonly adopted in industry. Recent work has shown that floating-point neural networks that have been verified to be robust can become vulnerable to adversarial attacks after quantization, and certification of the quantized representation is necessary to guarantee robustness. In this work, we present quantization-aware interval bound propagation (QA-IBP), a novel method for training robust QNNs. Inspired by advances in robust learning of non-quantized networks, our training algorithm computes the gradient of an abstract representation of the actual network. Unlike existing approaches, our method can handle the discrete semantics of QNNs. Based on QA-IBP, we also develop a complete verification procedure for verifying the adversarial robustness of QNNs, which is guaranteed to terminate and produce a correct answer. Compared to existing approaches, the key advantage of our verification procedure is that it runs entirely on GPU or other accelerator devices. We demonstrate experimentally that our approach significantly outperforms existing methods and establish the new state-of-the-art for training and certifying the robustness of QNNs.}, author = {Lechner, Mathias and Zikelic, Dorde and Chatterjee, Krishnendu and Henzinger, Thomas A and Rus, Daniela}, booktitle = {Proceedings of the 37th AAAI Conference on Artificial Intelligence}, isbn = {9781577358800}, location = {Washington, DC, United States}, number = {12}, pages = {14964--14973}, publisher = {Association for the Advancement of Artificial Intelligence}, title = {{Quantization-aware interval bound propagation for training certifiably robust quantized neural networks}}, doi = {10.1609/aaai.v37i12.26747}, volume = {37}, year = {2023}, } @inproceedings{13310, abstract = {Machine-learned systems are in widespread use for making decisions about humans, and it is important that they are fair, i.e., not biased against individuals based on sensitive attributes. We present runtime verification of algorithmic fairness for systems whose models are unknown, but are assumed to have a Markov chain structure. We introduce a specification language that can model many common algorithmic fairness properties, such as demographic parity, equal opportunity, and social burden. We build monitors that observe a long sequence of events as generated by a given system, and output, after each observation, a quantitative estimate of how fair or biased the system was on that run until that point in time. The estimate is proven to be correct modulo a variable error bound and a given confidence level, where the error bound gets tighter as the observed sequence gets longer. Our monitors are of two types, and use, respectively, frequentist and Bayesian statistical inference techniques. While the frequentist monitors compute estimates that are objectively correct with respect to the ground truth, the Bayesian monitors compute estimates that are correct subject to a given prior belief about the system’s model. Using a prototype implementation, we show how we can monitor if a bank is fair in giving loans to applicants from different social backgrounds, and if a college is fair in admitting students while maintaining a reasonable financial burden on the society. Although they exhibit different theoretical complexities in certain cases, in our experiments, both frequentist and Bayesian monitors took less than a millisecond to update their verdicts after each observation.}, author = {Henzinger, Thomas A and Karimi, Mahyar and Kueffner, Konstantin and Mallik, Kaushik}, booktitle = {Computer Aided Verification}, isbn = {9783031377020}, issn = {1611-3349}, location = {Paris, France}, pages = {358–382}, publisher = {Springer Nature}, title = {{Monitoring algorithmic fairness}}, doi = {10.1007/978-3-031-37703-7_17}, volume = {13965}, year = {2023}, } @inproceedings{13221, abstract = {The safety-liveness dichotomy is a fundamental concept in formal languages which plays a key role in verification. Recently, this dichotomy has been lifted to quantitative properties, which are arbitrary functions from infinite words to partially-ordered domains. We look into harnessing the dichotomy for the specific classes of quantitative properties expressed by quantitative automata. These automata contain finitely many states and rational-valued transition weights, and their common value functions Inf, Sup, LimInf, LimSup, LimInfAvg, LimSupAvg, and DSum map infinite words into the totallyordered domain of real numbers. In this automata-theoretic setting, we establish a connection between quantitative safety and topological continuity and provide an alternative characterization of quantitative safety and liveness in terms of their boolean counterparts. For all common value functions, we show how the safety closure of a quantitative automaton can be constructed in PTime, and we provide PSpace-complete checks of whether a given quantitative automaton is safe or live, with the exception of LimInfAvg and LimSupAvg automata, for which the safety check is in ExpSpace. Moreover, for deterministic Sup, LimInf, and LimSup automata, we give PTime decompositions into safe and live automata. These decompositions enable the separation of techniques for safety and liveness verification for quantitative specifications.}, author = {Boker, Udi and Henzinger, Thomas A and Mazzocchi, Nicolas Adrien and Sarac, Naci E}, booktitle = {34th International Conference on Concurrency Theory}, isbn = {9783959772990}, issn = {1868-8969}, location = {Antwerp, Belgium}, publisher = {Schloss Dagstuhl - Leibniz-Zentrum für Informatik}, title = {{Safety and liveness of quantitative automata}}, doi = {10.4230/LIPIcs.CONCUR.2023.17}, volume = {279}, year = {2023}, } @inproceedings{14405, abstract = {We introduce hypernode automata as a new specification formalism for hyperproperties of concurrent systems. They are finite automata with nodes labeled with hypernode logic formulas and transitions labeled with actions. A hypernode logic formula specifies relations between sequences of variable values in different system executions. Unlike HyperLTL, hypernode logic takes an asynchronous view on execution traces by constraining the values and the order of value changes of each variable without correlating the timing of the changes. Different execution traces are synchronized solely through the transitions of hypernode automata. Hypernode automata naturally combine asynchronicity at the node level with synchronicity at the transition level. We show that the model-checking problem for hypernode automata is decidable over action-labeled Kripke structures, whose actions induce transitions of the specification automata. For this reason, hypernode automaton is a suitable formalism for specifying and verifying asynchronous hyperproperties, such as declassifying observational determinism in multi-threaded programs.}, author = {Bartocci, Ezio and Henzinger, Thomas A and Nickovic, Dejan and Oliveira da Costa, Ana}, booktitle = {34th International Conference on Concurrency Theory}, isbn = {9783959772990}, issn = {18688969}, location = {Antwerp, Belgium}, publisher = {Schloss Dagstuhl - Leibniz-Zentrum für Informatik}, title = {{Hypernode automata}}, doi = {10.4230/LIPIcs.CONCUR.2023.21}, volume = {279}, year = {2023}, } @inproceedings{14454, abstract = {As AI and machine-learned software are used increasingly for making decisions that affect humans, it is imperative that they remain fair and unbiased in their decisions. To complement design-time bias mitigation measures, runtime verification techniques have been introduced recently to monitor the algorithmic fairness of deployed systems. Previous monitoring techniques assume full observability of the states of the (unknown) monitored system. Moreover, they can monitor only fairness properties that are specified as arithmetic expressions over the probabilities of different events. In this work, we extend fairness monitoring to systems modeled as partially observed Markov chains (POMC), and to specifications containing arithmetic expressions over the expected values of numerical functions on event sequences. The only assumptions we make are that the underlying POMC is aperiodic and starts in the stationary distribution, with a bound on its mixing time being known. These assumptions enable us to estimate a given property for the entire distribution of possible executions of the monitored POMC, by observing only a single execution. Our monitors observe a long run of the system and, after each new observation, output updated PAC-estimates of how fair or biased the system is. The monitors are computationally lightweight and, using a prototype implementation, we demonstrate their effectiveness on several real-world examples.}, author = {Henzinger, Thomas A and Kueffner, Konstantin and Mallik, Kaushik}, booktitle = {23rd International Conference on Runtime Verification}, isbn = {9783031442667}, issn = {1611-3349}, location = {Thessaloniki, Greece}, pages = {291--311}, publisher = {Springer Nature}, title = {{Monitoring algorithmic fairness under partial observations}}, doi = {10.1007/978-3-031-44267-4_15}, volume = {14245}, year = {2023}, } @inproceedings{14559, abstract = {We consider the problem of learning control policies in discrete-time stochastic systems which guarantee that the system stabilizes within some specified stabilization region with probability 1. Our approach is based on the novel notion of stabilizing ranking supermartingales (sRSMs) that we introduce in this work. Our sRSMs overcome the limitation of methods proposed in previous works whose applicability is restricted to systems in which the stabilizing region cannot be left once entered under any control policy. We present a learning procedure that learns a control policy together with an sRSM that formally certifies probability 1 stability, both learned as neural networks. We show that this procedure can also be adapted to formally verifying that, under a given Lipschitz continuous control policy, the stochastic system stabilizes within some stabilizing region with probability 1. Our experimental evaluation shows that our learning procedure can successfully learn provably stabilizing policies in practice.}, author = {Ansaripour, Matin and Chatterjee, Krishnendu and Henzinger, Thomas A and Lechner, Mathias and Zikelic, Dorde}, booktitle = {21st International Symposium on Automated Technology for Verification and Analysis}, isbn = {9783031453281}, issn = {1611-3349}, location = {Singapore, Singapore}, pages = {357--379}, publisher = {Springer Nature}, title = {{Learning provably stabilizing neural controllers for discrete-time stochastic systems}}, doi = {10.1007/978-3-031-45329-8_17}, volume = {14215}, year = {2023}, } @inproceedings{13228, abstract = {A machine-learned system that is fair in static decision-making tasks may have biased societal impacts in the long-run. This may happen when the system interacts with humans and feedback patterns emerge, reinforcing old biases in the system and creating new biases. While existing works try to identify and mitigate long-run biases through smart system design, we introduce techniques for monitoring fairness in real time. Our goal is to build and deploy a monitor that will continuously observe a long sequence of events generated by the system in the wild, and will output, with each event, a verdict on how fair the system is at the current point in time. The advantages of monitoring are two-fold. Firstly, fairness is evaluated at run-time, which is important because unfair behaviors may not be eliminated a priori, at design-time, due to partial knowledge about the system and the environment, as well as uncertainties and dynamic changes in the system and the environment, such as the unpredictability of human behavior. Secondly, monitors are by design oblivious to how the monitored system is constructed, which makes them suitable to be used as trusted third-party fairness watchdogs. They function as computationally lightweight statistical estimators, and their correctness proofs rely on the rigorous analysis of the stochastic process that models the assumptions about the underlying dynamics of the system. We show, both in theory and experiments, how monitors can warn us (1) if a bank’s credit policy over time has created an unfair distribution of credit scores among the population, and (2) if a resource allocator’s allocation policy over time has made unfair allocations. Our experiments demonstrate that the monitors introduce very low overhead. We believe that runtime monitoring is an important and mathematically rigorous new addition to the fairness toolbox.}, author = {Henzinger, Thomas A and Karimi, Mahyar and Kueffner, Konstantin and Mallik, Kaushik}, booktitle = {FAccT '23: Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency}, isbn = {9781450372527}, location = {Chicago, IL, United States}, pages = {604--614}, publisher = {Association for Computing Machinery}, title = {{Runtime monitoring of dynamic fairness properties}}, doi = {10.1145/3593013.3594028}, year = {2023}, } @article{13263, abstract = {Motivation: Boolean networks are simple but efficient mathematical formalism for modelling complex biological systems. However, having only two levels of activation is sometimes not enough to fully capture the dynamics of real-world biological systems. Hence, the need for multi-valued networks (MVNs), a generalization of Boolean networks. Despite the importance of MVNs for modelling biological systems, only limited progress has been made on developing theories, analysis methods, and tools that can support them. In particular, the recent use of trap spaces in Boolean networks made a great impact on the field of systems biology, but there has been no similar concept defined and studied for MVNs to date. Results: In this work, we generalize the concept of trap spaces in Boolean networks to that in MVNs. We then develop the theory and the analysis methods for trap spaces in MVNs. In particular, we implement all proposed methods in a Python package called trapmvn. Not only showing the applicability of our approach via a realistic case study, we also evaluate the time efficiency of the method on a large collection of real-world models. The experimental results confirm the time efficiency, which we believe enables more accurate analysis on larger and more complex multi-valued models.}, author = {Trinh, Van Giang and Benhamou, Belaid and Henzinger, Thomas A and Pastva, Samuel}, issn = {1367-4811}, journal = {Bioinformatics}, number = {Supplement_1}, pages = {i513--i522}, publisher = {Oxford Academic}, title = {{Trap spaces of multi-valued networks: Definition, computation, and applications}}, doi = {10.1093/bioinformatics/btad262}, volume = {39}, year = {2023}, } @inproceedings{14718, abstract = {Binary decision diagrams (BDDs) are one of the fundamental data structures in formal methods and computer science in general. However, the performance of BDD-based algorithms greatly depends on memory latency due to the reliance on large hash tables and thus, by extension, on the speed of random memory access. This hinders the full utilisation of resources available on modern CPUs, since the absolute memory latency has not improved significantly for at least a decade. In this paper, we explore several implementation techniques that improve the performance of BDD manipulation either through enhanced memory locality or by partially eliminating random memory access. On a benchmark suite of 600+ BDDs derived from real-world applications, we demonstrate runtime that is comparable or better than parallelising the same operations on eight CPU cores. }, author = {Pastva, Samuel and Henzinger, Thomas A}, booktitle = {Proceedings of the 23rd Conference on Formal Methods in Computer-Aided Design}, isbn = {9783854480600}, location = {Ames, IA, United States}, pages = {122--131}, publisher = {TU Vienna Academic Press}, title = {{Binary decision diagrams on modern hardware}}, doi = {10.34727/2023/isbn.978-3-85448-060-0_20}, year = {2023}, } @inproceedings{14830, abstract = {We study the problem of learning controllers for discrete-time non-linear stochastic dynamical systems with formal reach-avoid guarantees. This work presents the first method for providing formal reach-avoid guarantees, which combine and generalize stability and safety guarantees, with a tolerable probability threshold p in [0,1] over the infinite time horizon. Our method leverages advances in machine learning literature and it represents formal certificates as neural networks. In particular, we learn a certificate in the form of a reach-avoid supermartingale (RASM), a novel notion that we introduce in this work. Our RASMs provide reachability and avoidance guarantees by imposing constraints on what can be viewed as a stochastic extension of level sets of Lyapunov functions for deterministic systems. Our approach solves several important problems -- it can be used to learn a control policy from scratch, to verify a reach-avoid specification for a fixed control policy, or to fine-tune a pre-trained policy if it does not satisfy the reach-avoid specification. We validate our approach on 3 stochastic non-linear reinforcement learning tasks.}, author = {Zikelic, Dorde and Lechner, Mathias and Henzinger, Thomas A and Chatterjee, Krishnendu}, booktitle = {Proceedings of the 37th AAAI Conference on Artificial Intelligence}, issn = {2374-3468}, keywords = {General Medicine}, location = {Washington, DC, United States}, number = {10}, pages = {11926--11935}, publisher = {Association for the Advancement of Artificial Intelligence}, title = {{Learning control policies for stochastic systems with reach-avoid guarantees}}, doi = {10.1609/aaai.v37i10.26407}, volume = {37}, year = {2023}, } @article{13234, abstract = {Neural-network classifiers achieve high accuracy when predicting the class of an input that they were trained to identify. Maintaining this accuracy in dynamic environments, where inputs frequently fall outside the fixed set of initially known classes, remains a challenge. We consider the problem of monitoring the classification decisions of neural networks in the presence of novel classes. For this purpose, we generalize our recently proposed abstraction-based monitor from binary output to real-valued quantitative output. This quantitative output enables new applications, two of which we investigate in the paper. As our first application, we introduce an algorithmic framework for active monitoring of a neural network, which allows us to learn new classes dynamically and yet maintain high monitoring performance. As our second application, we present an offline procedure to retrain the neural network to improve the monitor’s detection performance without deteriorating the network’s classification accuracy. Our experimental evaluation demonstrates both the benefits of our active monitoring framework in dynamic scenarios and the effectiveness of the retraining procedure.}, author = {Kueffner, Konstantin and Lukina, Anna and Schilling, Christian and Henzinger, Thomas A}, issn = {1433-2787}, journal = {International Journal on Software Tools for Technology Transfer}, pages = {575--592}, publisher = {Springer Nature}, title = {{Into the unknown: Active monitoring of neural networks (extended version)}}, doi = {10.1007/s10009-023-00711-4}, volume = {25}, year = {2023}, } @inproceedings{15023, abstract = {Reinforcement learning has shown promising results in learning neural network policies for complicated control tasks. However, the lack of formal guarantees about the behavior of such policies remains an impediment to their deployment. We propose a novel method for learning a composition of neural network policies in stochastic environments, along with a formal certificate which guarantees that a specification over the policy's behavior is satisfied with the desired probability. Unlike prior work on verifiable RL, our approach leverages the compositional nature of logical specifications provided in SpectRL, to learn over graphs of probabilistic reach-avoid specifications. The formal guarantees are provided by learning neural network policies together with reach-avoid supermartingales (RASM) for the graph’s sub-tasks and then composing them into a global policy. We also derive a tighter lower bound compared to previous work on the probability of reach-avoidance implied by a RASM, which is required to find a compositional policy with an acceptable probabilistic threshold for complex tasks with multiple edge policies. We implement a prototype of our approach and evaluate it on a Stochastic Nine Rooms environment.}, author = {Zikelic, Dorde and Lechner, Mathias and Verma, Abhinav and Chatterjee, Krishnendu and Henzinger, Thomas A}, booktitle = {37th Conference on Neural Information Processing Systems}, location = {New Orleans, LO, United States}, title = {{Compositional policy learning in stochastic control systems with formal guarantees}}, year = {2023}, } @inproceedings{14076, abstract = {Hyperproperties are properties that relate multiple execution traces. Previous work on monitoring hyperproperties focused on synchronous hyperproperties, usually specified in HyperLTL. When monitoring synchronous hyperproperties, all traces are assumed to proceed at the same speed. We introduce (multi-trace) prefix transducers and show how to use them for monitoring synchronous as well as, for the first time, asynchronous hyperproperties. Prefix transducers map multiple input traces into one or more output traces by incrementally matching prefixes of the input traces against expressions similar to regular expressions. The prefixes of different traces which are consumed by a single matching step of the monitor may have different lengths. The deterministic and executable nature of prefix transducers makes them more suitable as an intermediate formalism for runtime verification than logical specifications, which tend to be highly non-deterministic, especially in the case of asynchronous hyperproperties. We report on a set of experiments about monitoring asynchronous version of observational determinism.}, author = {Chalupa, Marek and Henzinger, Thomas A}, booktitle = {23nd International Conference on Runtime Verification}, isbn = {978-3-031-44266-7}, location = {Thessaloniki, Greek}, pages = {168--190}, publisher = {Springer Nature}, title = {{Monitoring hyperproperties with prefix transducers}}, doi = {10.1007/978-3-031-44267-4_9}, volume = {14245}, year = {2023}, }