TY - CONF AB - We study the problem of specifying sequential information-flow properties of systems. Information-flow properties are hyperproperties, as they compare different traces of a system. Sequential information-flow properties can express changes, over time, in the information-flow constraints. For example, information-flow constraints during an initialization phase of a system may be different from information-flow constraints that are required during the operation phase. We formalize several variants of interpreting sequential information-flow constraints, which arise from different assumptions about what can be observed of the system. For this purpose, we introduce a first-order logic, called Hypertrace Logic, with both trace and time quantifiers for specifying linear-time hyperproperties. We prove that HyperLTL, which corresponds to a fragment of Hypertrace Logic with restricted quantifier prefixes, cannot specify the majority of the studied variants of sequential information flow, including all variants in which the transition between sequential phases (such as initialization and operation) happens asynchronously. Our results rely on new equivalences between sets of traces that cannot be distinguished by certain classes of formulas from Hypertrace Logic. This presents a new approach to proving inexpressiveness results for HyperLTL. AU - Bartocci, Ezio AU - Ferrere, Thomas AU - Henzinger, Thomas A AU - Nickovic, Dejan AU - Da Costa, Ana Oliveira ID - 10774 SN - 03029743 T2 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) TI - Flavors of sequential information flow VL - 13182 ER - TY - CONF AB - World models learn behaviors in a latent imagination space to enhance the sample-efficiency of deep reinforcement learning (RL) algorithms. While learning world models for high-dimensional observations (e.g., pixel inputs) has become practicable on standard RL benchmarks and some games, their effectiveness in real-world robotics applications has not been explored. In this paper, we investigate how such agents generalize to real-world autonomous vehicle control tasks, where advanced model-free deep RL algorithms fail. In particular, we set up a series of time-lap tasks for an F1TENTH racing robot, equipped with a high-dimensional LiDAR sensor, on a set of test tracks with a gradual increase in their complexity. In this continuous-control setting, we show that model-based agents capable of learning in imagination substantially outperform model-free agents with respect to performance, sample efficiency, successful task completion, and generalization. Moreover, we show that the generalization ability of model-based agents strongly depends on the choice of their observation model. We provide extensive empirical evidence for the effectiveness of world models provided with long enough memory horizons in sim2real tasks. AU - Brunnbauer, Axel AU - Berducci, Luigi AU - Brandstatter, Andreas AU - Lechner, Mathias AU - Hasani, Ramin AU - Rus, Daniela AU - Grosu, Radu ID - 12010 SN - 1050-4729 T2 - 2022 International Conference on Robotics and Automation TI - Latent imagination facilitates zero-shot transfer in autonomous racing ER - TY - CONF AB - We propose an algorithmic approach for synthesizing linear hybrid automata from time-series data. Unlike existing approaches, our approach provides a whole family of models with the same discrete structure but different dynamics. Each model in the family is guaranteed to capture the input data up to a precision error ε, in the following sense: For each time series, the model contains an execution that is ε-close to the data points. Our construction allows to effectively choose a model from this family with minimal precision error ε. We demonstrate the algorithm’s efficiency and its ability to find precise models in two case studies. AU - Garcia Soto, Miriam AU - Henzinger, Thomas A AU - Schilling, Christian ID - 12171 SN - 0302-9743 T2 - 20th International Symposium on Automated Technology for Verification and Analysis TI - Synthesis of parametric hybrid automata from time series VL - 13505 ER - TY - CONF AB - We explore the notion of history-determinism in the context of timed automata (TA). History-deterministic automata are those in which nondeterminism can be resolved on the fly, based on the run constructed thus far. History-determinism is a robust property that admits different game-based characterisations, and history-deterministic specifications allow for game-based verification without an expensive determinization step. We show yet another characterisation of history-determinism in terms of fair simulation, at the general level of labelled transition systems: a system is history-deterministic precisely if and only if it fairly simulates all language smaller systems. For timed automata over infinite timed words it is known that universality is undecidable for Büchi TA. We show that for history-deterministic TA with arbitrary parity acceptance, timed universality, inclusion, and synthesis all remain decidable and are ExpTime-complete. For the subclass of TA with safety or reachability acceptance, we show that checking whether such an automaton is history-deterministic is decidable (in ExpTime), and history-deterministic TA with safety acceptance are effectively determinizable without introducing new automata states. AU - Henzinger, Thomas A AU - Lehtinen, Karoliina AU - Totzke, Patrick ID - 12508 SN - 1868-8969 T2 - 33rd International Conference on Concurrency Theory TI - History-deterministic timed automata VL - 243 ER - TY - CONF AB - A graph game is a two-player zero-sum game in which the players move a token throughout a graph to produce an infinite path, which determines the winner or payoff of the game. In bidding games, both players have budgets, and in each turn, we hold an "auction" (bidding) to determine which player moves the token. In this survey, we consider several bidding mechanisms and their effect on the properties of the game. Specifically, bidding games, and in particular bidding games of infinite duration, have an intriguing equivalence with random-turn games in which in each turn, the player who moves is chosen randomly. We summarize how minor changes in the bidding mechanism lead to unexpected differences in the equivalence with random-turn games. AU - Avni, Guy AU - Henzinger, Thomas A ID - 12509 SN - 1868-8969 T2 - 47th International Symposium on Mathematical Foundations of Computer Science TI - An updated survey of bidding games on graphs VL - 241 ER - TY - GEN AB - 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 can make adversarial training suitable for real-world robot applications. We evaluate a wide variety of robot learning tasks ranging from autonomous driving in a high-fidelity environment amenable to sim-to-real deployment, to mobile robot gesture recognition. Our results demonstrate that, while these techniques make incremental improvements on the trade-off on a relative scale, the negative side-effects caused by adversarial training still outweigh the improvements by an order of magnitude. We conclude that more substantial advances in robust learning methods are necessary before they can benefit robot learning tasks in practice. AU - Lechner, Mathias AU - Amini, Alexander AU - Rus, Daniela AU - Henzinger, Thomas A ID - 11366 T2 - arXiv TI - Revisiting the adversarial robustness-accuracy tradeoff in robot learning ER - TY - CONF AB - We present a formal framework for the online black-box monitoring of software using monitors with quantitative verdict functions. Quantitative verdict functions have several advantages. First, quantitative monitors can be approximate, i.e., the value of the verdict function does not need to correspond exactly to the value of the property under observation. Second, quantitative monitors can be quantified universally, i.e., for every possible observed behavior, the monitor tries to make the best effort to estimate the value of the property under observation. Third, quantitative monitors can watch boolean as well as quantitative properties, such as average response time. Fourth, quantitative monitors can use non-finite-state resources, such as counters. As a consequence, quantitative monitors can be compared according to how many resources they use (e.g., the number of counters) and how precisely they approximate the property under observation. This allows for a rich spectrum of cost-precision trade-offs in monitoring software. AU - Henzinger, Thomas A ID - 10891 SN - 0302-9743 T2 - Software Verification TI - Quantitative monitoring of software VL - 13124 ER - TY - CONF AB - Contract-based design is a promising methodology for taming the complexity of developing sophisticated systems. A formal contract distinguishes between assumptions, which are constraints that the designer of a component puts on the environments in which the component can be used safely, and guarantees, which are promises that the designer asks from the team that implements the component. A theory of formal contracts can be formalized as an interface theory, which supports the composition and refinement of both assumptions and guarantees. Although there is a rich landscape of contract-based design methods that address functional and extra-functional properties, we present the first interface theory that is designed for ensuring system-wide security properties. Our framework provides a refinement relation and a composition operation that support both incremental design and independent implementability. We develop our theory for both stateless and stateful interfaces. We illustrate the applicability of our framework with an example inspired from the automotive domain. AU - Bartocci, Ezio AU - Ferrere, Thomas AU - Henzinger, Thomas A AU - Nickovic, Dejan AU - Da Costa, Ana Oliveira ID - 11355 SN - 0302-9743 T2 - Fundamental Approaches to Software Engineering TI - Information-flow interfaces VL - 13241 ER - TY - CONF AB - Quantitative monitoring can be universal and approximate: For every finite sequence of observations, the specification provides a value and the monitor outputs a best-effort approximation of it. The quality of the approximation may depend on the resources that are available to the monitor. By taking to the limit the sequences of specification values and monitor outputs, we obtain precision-resource trade-offs also for limit monitoring. This paper provides a formal framework for studying such trade-offs using an abstract interpretation for monitors: For each natural number n, the aggregate semantics of a monitor at time n is an equivalence relation over all sequences of at most n observations so that two equivalent sequences are indistinguishable to the monitor and thus mapped to the same output. This abstract interpretation of quantitative monitors allows us to measure the number of equivalence classes (or “resource use”) that is necessary for a certain precision up to a certain time, or at any time. Our framework offers several insights. For example, we identify a family of specifications for which any resource-optimal exact limit monitor is independent of any error permitted over finite traces. Moreover, we present a specification for which any resource-optimal approximate limit monitor does not minimize its resource use at any time. AU - Henzinger, Thomas A AU - Mazzocchi, Nicolas Adrien AU - Sarac, Naci E ID - 11775 SN - 0302-9743 T2 - 22nd International Conference on Runtime Verification TI - Abstract monitors for quantitative specifications VL - 13498 ER - TY - JOUR AB - Continuous-time neural networks are a class of machine learning systems that can tackle representation learning on spatiotemporal decision-making tasks. These models are typically represented by continuous differential equations. However, their expressive power when they are deployed on computers is bottlenecked by numerical differential equation solvers. This limitation has notably slowed down the scaling and understanding of numerous natural physical phenomena such as the dynamics of nervous systems. Ideally, we would circumvent this bottleneck by solving the given dynamical system in closed form. This is known to be intractable in general. Here, we show that it is possible to closely approximate the interaction between neurons and synapses—the building blocks of natural and artificial neural networks—constructed by liquid time-constant networks efficiently in closed form. To this end, we compute a tightly bounded approximation of the solution of an integral appearing in liquid time-constant dynamics that has had no known closed-form solution so far. This closed-form solution impacts the design of continuous-time and continuous-depth neural models. For instance, since time appears explicitly in closed form, the formulation relaxes the need for complex numerical solvers. Consequently, we obtain models that are between one and five orders of magnitude faster in training and inference compared with differential equation-based counterparts. More importantly, in contrast to ordinary differential equation-based continuous networks, closed-form networks can scale remarkably well compared with other deep learning instances. Lastly, as these models are derived from liquid networks, they show good performance in time-series modelling compared with advanced recurrent neural network models. AU - Hasani, Ramin AU - Lechner, Mathias AU - Amini, Alexander AU - Liebenwein, Lucas AU - Ray, Aaron AU - Tschaikowski, Max AU - Teschl, Gerald AU - Rus, Daniela ID - 12147 IS - 11 JF - Nature Machine Intelligence KW - Artificial Intelligence KW - Computer Networks and Communications KW - Computer Vision and Pattern Recognition KW - Human-Computer Interaction KW - Software SN - 2522-5839 TI - Closed-form continuous-time neural networks VL - 4 ER - TY - THES AB - Deep learning has enabled breakthroughs in challenging computing problems and has emerged as the standard problem-solving tool for computer vision and natural language processing tasks. One exception to this trend is safety-critical tasks where robustness and resilience requirements contradict the black-box nature of neural networks. To deploy deep learning methods for these tasks, it is vital to provide guarantees on neural network agents' safety and robustness criteria. This can be achieved by developing formal verification methods to verify the safety and robustness properties of neural networks. Our goal is to design, develop and assess safety verification methods for neural networks to improve their reliability and trustworthiness in real-world applications. This thesis establishes techniques for the verification of compressed and adversarially trained models as well as the design of novel neural networks for verifiably safe decision-making. First, we establish the problem of verifying quantized neural networks. Quantization is a technique that trades numerical precision for the computational efficiency of running a neural network and is widely adopted in industry. We show that neglecting the reduced precision when verifying a neural network can lead to wrong conclusions about the robustness and safety of the network, highlighting that novel techniques for quantized network verification are necessary. We introduce several bit-exact verification methods explicitly designed for quantized neural networks and experimentally confirm on realistic networks that the network's robustness and other formal properties are affected by the quantization. Furthermore, we perform a case study providing evidence that adversarial training, a standard technique for making neural networks more robust, has detrimental effects on the network's performance. This robustness-accuracy tradeoff has been studied before regarding the accuracy obtained on classification datasets where each data point is independent of all other data points. On the other hand, we investigate the tradeoff empirically in robot learning settings where a both, a high accuracy and a high robustness, are desirable. Our results suggest that the negative side-effects of adversarial training outweigh its robustness benefits in practice. Finally, we consider the problem of verifying safety when running a Bayesian neural network policy in a feedback loop with systems over the infinite time horizon. Bayesian neural networks are probabilistic models for learning uncertainties in the data and are therefore often used on robotic and healthcare applications where data is inherently stochastic. We introduce a method for recalibrating Bayesian neural networks so that they yield probability distributions over safe decisions only. Our method learns a safety certificate that guarantees safety over the infinite time horizon to determine which decisions are safe in every possible state of the system. We demonstrate the effectiveness of our approach on a series of reinforcement learning benchmarks. AU - Lechner, Mathias ID - 11362 KW - neural networks KW - verification KW - machine learning SN - 978-3-99078-017-6 TI - Learning verifiable representations ER - TY - CONF AB - We propose a novel algorithm to decide the language inclusion between (nondeterministic) Büchi automata, a PSPACE-complete problem. Our approach, like others before, leverage a notion of quasiorder to prune the search for a counterexample by discarding candidates which are subsumed by others for the quasiorder. Discarded candidates are guaranteed to not compromise the completeness of the algorithm. The novelty of our work lies in the quasiorder used to discard candidates. We introduce FORQs (family of right quasiorders) that we obtain by adapting the notion of family of right congruences put forward by Maler and Staiger in 1993. We define a FORQ-based inclusion algorithm which we prove correct and instantiate it for a specific FORQ, called the structural FORQ, induced by the Büchi automaton to the right of the inclusion sign. The resulting implementation, called FORKLIFT, scales up better than the state-of-the-art on a variety of benchmarks including benchmarks from program verification and theorem proving for word combinatorics. Artifact: https://doi.org/10.5281/zenodo.6552870 AU - Doveri, Kyveli AU - Ganty, Pierre AU - Mazzocchi, Nicolas Adrien ID - 12302 SN - 0302-9743 T2 - Computer Aided Verification TI - FORQ-based language inclusion formal testing VL - 13372 ER - TY - CONF AB - An automaton is history-deterministic (HD) if one can safely resolve its non-deterministic choices on the fly. In a recent paper, Henzinger, Lehtinen and Totzke studied this in the context of Timed Automata [9], where it was conjectured that the class of timed ω-languages recognised by HD-timed automata strictly extends that of deterministic ones. We provide a proof for this fact. AU - Bose, Sougata AU - Henzinger, Thomas A AU - Lehtinen, Karoliina AU - Schewe, Sven AU - Totzke, Patrick ID - 12175 SN - 0302-9743 T2 - 16th International Conference on Reachability Problems TI - History-deterministic timed automata are not determinizable VL - 13608 ER - TY - JOUR AB - We introduce a new statistical verification algorithm that formally quantifies the behavioral robustness of any time-continuous process formulated as a continuous-depth model. Our algorithm solves a set of global optimization (Go) problems over a given time horizon to construct a tight enclosure (Tube) of the set of all process executions starting from a ball of initial states. We call our algorithm GoTube. Through its construction, GoTube ensures that the bounding tube is conservative up to a desired probability and up to a desired tightness. GoTube is implemented in JAX and optimized to scale to complex continuous-depth neural network models. Compared to advanced reachability analysis tools for time-continuous neural networks, GoTube does not accumulate overapproximation errors between time steps and avoids the infamous wrapping effect inherent in symbolic techniques. We show that GoTube substantially outperforms state-of-the-art verification tools in terms of the size of the initial ball, speed, time-horizon, task completion, and scalability on a large set of experiments. GoTube is stable and sets the state-of-the-art in terms of its ability to scale to time horizons well beyond what has been previously possible. AU - Gruenbacher, Sophie A. AU - Lechner, Mathias AU - Hasani, Ramin AU - Rus, Daniela AU - Henzinger, Thomas A AU - Smolka, Scott A. AU - Grosu, Radu ID - 12510 IS - 6 JF - Proceedings of the AAAI Conference on Artificial Intelligence KW - General Medicine SN - 2159-5399 TI - GoTube: Scalable statistical verification of continuous-depth models VL - 36 ER - TY - JOUR AB - We consider the problem of formally verifying almost-sure (a.s.) asymptotic stability in discrete-time nonlinear stochastic control systems. While verifying stability in deterministic control systems is extensively studied in the literature, verifying stability in stochastic control systems is an open problem. The few existing works on this topic either consider only specialized forms of stochasticity or make restrictive assumptions on the system, rendering them inapplicable to learning algorithms with neural network policies. In this work, we present an approach for general nonlinear stochastic control problems with two novel aspects: (a) instead of classical stochastic extensions of Lyapunov functions, we use ranking supermartingales (RSMs) to certify a.s. asymptotic stability, and (b) we present a method for learning neural network RSMs. We prove that our approach guarantees a.s. asymptotic stability of the system and provides the first method to obtain bounds on the stabilization time, which stochastic Lyapunov functions do not. Finally, we validate our approach experimentally on a set of nonlinear stochastic reinforcement learning environments with neural network policies. AU - Lechner, Mathias AU - Zikelic, Dorde AU - Chatterjee, Krishnendu AU - Henzinger, Thomas A ID - 12511 IS - 7 JF - Proceedings of the AAAI Conference on Artificial Intelligence KW - General Medicine SN - 2159-5399 TI - Stability verification in stochastic control systems via neural network supermartingales VL - 36 ER - TY - GEN AB - In this work, we address the problem of learning provably stable neural network policies for stochastic control systems. While recent work has demonstrated the feasibility of certifying given policies using martingale theory, the problem of how to learn such policies is little explored. Here, we study the effectiveness of jointly learning a policy together with a martingale certificate that proves its stability using a single learning algorithm. We observe that the joint optimization problem becomes easily stuck in local minima when starting from a randomly initialized policy. Our results suggest that some form of pre-training of the policy is required for the joint optimization to repair and verify the policy successfully. AU - Zikelic, Dorde AU - Lechner, Mathias AU - Chatterjee, Krishnendu AU - Henzinger, Thomas A ID - 14601 T2 - arXiv TI - Learning stabilizing policies in stochastic control systems ER - TY - GEN AB - 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. AU - Zikelic, Dorde AU - Lechner, Mathias AU - Henzinger, Thomas A AU - Chatterjee, Krishnendu ID - 14600 T2 - arXiv TI - Learning control policies for stochastic systems with reach-avoid guarantees ER - TY - JOUR AB - Gradual typing is a principled means for mixing typed and untyped code. But typed and untyped code often exhibit different programming patterns. There is already substantial research investigating gradually giving types to code exhibiting typical untyped patterns, and some research investigating gradually removing types from code exhibiting typical typed patterns. This paper investigates how to extend these established gradual-typing concepts to give formal guarantees not only about how to change types as code evolves but also about how to change such programming patterns as well. In particular, we explore mixing untyped "structural" code with typed "nominal" code in an object-oriented language. But whereas previous work only allowed "nominal" objects to be treated as "structural" objects, we also allow "structural" objects to dynamically acquire certain nominal types, namely interfaces. We present a calculus that supports such "cross-paradigm" code migration and interoperation in a manner satisfying both the static and dynamic gradual guarantees, and demonstrate that the calculus can be implemented efficiently. AU - Mühlböck, Fabian AU - Tate, Ross ID - 10153 JF - Proceedings of the ACM on Programming Languages KW - gradual typing KW - gradual guarantee KW - nominal KW - structural KW - call tags TI - Transitioning from structural to nominal code with efficient gradual typing VL - 5 ER - TY - CONF AB - We show that Neural ODEs, an emerging class of timecontinuous neural networks, can be verified by solving a set of global-optimization problems. For this purpose, we introduce Stochastic Lagrangian Reachability (SLR), an abstraction-based technique for constructing a tight Reachtube (an over-approximation of the set of reachable states over a given time-horizon), and provide stochastic guarantees in the form of confidence intervals for the Reachtube bounds. SLR inherently avoids the infamous wrapping effect (accumulation of over-approximation errors) by performing local optimization steps to expand safe regions instead of repeatedly forward-propagating them as is done by deterministic reachability methods. To enable fast local optimizations, we introduce a novel forward-mode adjoint sensitivity method to compute gradients without the need for backpropagation. Finally, we establish asymptotic and non-asymptotic convergence rates for SLR. AU - Grunbacher, Sophie AU - Hasani, Ramin AU - Lechner, Mathias AU - Cyranka, Jacek AU - Smolka, Scott A AU - Grosu, Radu ID - 10669 IS - 13 SN - 2159-5399 T2 - Proceedings of the AAAI Conference on Artificial Intelligence TI - On the verification of neural ODEs with stochastic guarantees VL - 35 ER - TY - CONF AB - We introduce a new class of time-continuous recurrent neural network models. Instead of declaring a learning system’s dynamics by implicit nonlinearities, we construct networks of linear first-order dynamical systems modulated via nonlinear interlinked gates. The resulting models represent dynamical systems with varying (i.e., liquid) time-constants coupled to their hidden state, with outputs being computed by numerical differential equation solvers. These neural networks exhibit stable and bounded behavior, yield superior expressivity within the family of neural ordinary differential equations, and give rise to improved performance on time-series prediction tasks. To demonstrate these properties, we first take a theoretical approach to find bounds over their dynamics, and compute their expressive power by the trajectory length measure in a latent trajectory space. We then conduct a series of time-series prediction experiments to manifest the approximation capability of Liquid Time-Constant Networks (LTCs) compared to classical and modern RNNs. AU - Hasani, Ramin AU - Lechner, Mathias AU - Amini, Alexander AU - Rus, Daniela AU - Grosu, Radu ID - 10671 IS - 9 SN - 2159-5399 T2 - Proceedings of the AAAI Conference on Artificial Intelligence TI - Liquid time-constant networks VL - 35 ER - TY - CONF AB - Robustness to variations in lighting conditions is a key objective for any deep vision system. To this end, our paper extends the receptive field of convolutional neural networks with two residual components, ubiquitous in the visual processing system of vertebrates: On-center and off-center pathways, with an excitatory center and inhibitory surround; OOCS for short. The On-center pathway is excited by the presence of a light stimulus in its center, but not in its surround, whereas the Off-center pathway is excited by the absence of a light stimulus in its center, but not in its surround. We design OOCS pathways via a difference of Gaussians, with their variance computed analytically from the size of the receptive fields. OOCS pathways complement each other in their response to light stimuli, ensuring this way a strong edge-detection capability, and as a result an accurate and robust inference under challenging lighting conditions. We provide extensive empirical evidence showing that networks supplied with OOCS pathways gain accuracy and illumination-robustness from the novel edge representation, compared to other baselines. AU - Babaiee, Zahra AU - Hasani, Ramin AU - Lechner, Mathias AU - Rus, Daniela AU - Grosu, Radu ID - 10668 SN - 2640-3498 T2 - Proceedings of the 38th International Conference on Machine Learning TI - On-off center-surround receptive fields for accurate and robust image classification VL - 139 ER - TY - CONF AB - Imitation learning enables high-fidelity, vision-based learning of policies within rich, photorealistic environments. However, such techniques often rely on traditional discrete-time neural models and face difficulties in generalizing to domain shifts by failing to account for the causal relationships between the agent and the environment. In this paper, we propose a theoretical and experimental framework for learning causal representations using continuous-time neural networks, specifically over their discrete-time counterparts. We evaluate our method in the context of visual-control learning of drones over a series of complex tasks, ranging from short- and long-term navigation, to chasing static and dynamic objects through photorealistic environments. Our results demonstrate that causal continuous-time deep models can perform robust navigation tasks, where advanced recurrent models fail. These models learn complex causal control representations directly from raw visual inputs and scale to solve a variety of tasks using imitation learning. AU - Vorbach, Charles J AU - Hasani, Ramin AU - Amini, Alexander AU - Lechner, Mathias AU - Rus, Daniela ID - 10670 T2 - 35th Conference on Neural Information Processing Systems TI - Causal navigation by continuous-time neural networks ER - TY - CONF AB - Civl is a static verifier for concurrent programs designed around the conceptual framework of layered refinement, which views the task of verifying a program as a sequence of program simplification steps each justified by its own invariant. Civl verifies a layered concurrent program that compactly expresses all the programs in this sequence and the supporting invariants. This paper presents the design and implementation of the Civl verifier. AU - Kragl, Bernhard AU - Qadeer, Shaz ED - Ruzica, Piskac ED - Whalen, Michael W. ID - 10688 SN - 978-3-85448-046-4 T2 - Proceedings of the 21st Conference on Formal Methods in Computer-Aided Design TI - The Civl verifier VL - 2 ER - TY - GEN AB - We comment on two formal proofs of Fermat's sum of two squares theorem, written using the Mathematical Components libraries of the Coq proof assistant. The first one follows Zagier's celebrated one-sentence proof; the second follows David Christopher's recent new proof relying on partition-theoretic arguments. Both formal proofs rely on a general property of involutions of finite sets, of independent interest. The proof technique consists for the most part of automating recurrent tasks (such as case distinctions and computations on natural numbers) via ad hoc tactics. AU - Dubach, Guillaume AU - Mühlböck, Fabian ID - 9281 T2 - arXiv TI - Formal verification of Zagier's one-sentence proof ER - TY - CONF AB - Formal verification of neural networks is an active topic of research, and recent advances have significantly increased the size of the networks that verification tools can handle. However, most methods are designed for verification of an idealized model of the actual network which works over real arithmetic and ignores rounding imprecisions. This idealization is in stark contrast to network quantization, which is a technique that trades numerical precision for computational efficiency and is, therefore, often applied in practice. Neglecting rounding errors of such low-bit quantized neural networks has been shown to lead to wrong conclusions about the network’s correctness. Thus, the desired approach for verifying quantized neural networks would be one that takes these rounding errors into account. In this paper, we show that verifying the bitexact implementation of quantized neural networks with bitvector specifications is PSPACE-hard, even though verifying idealized real-valued networks and satisfiability of bit-vector specifications alone are each in NP. Furthermore, we explore several practical heuristics toward closing the complexity gap between idealized and bit-exact verification. In particular, we propose three techniques for making SMT-based verification of quantized neural networks more scalable. Our experiments demonstrate that our proposed methods allow a speedup of up to three orders of magnitude over existing approaches. AU - Henzinger, Thomas A AU - Lechner, Mathias AU - Zikelic, Dorde ID - 10665 IS - 5A SN - 2159-5399 T2 - Proceedings of the AAAI Conference on Artificial Intelligence TI - Scalable verification of quantized neural networks VL - 35 ER - TY - CONF AB - Bayesian neural networks (BNNs) place distributions over the weights of a neural network to model uncertainty in the data and the network's prediction. We consider the problem of verifying safety when running a Bayesian neural network policy in a feedback loop with infinite time horizon systems. Compared to the existing sampling-based approaches, which are inapplicable to the infinite time horizon setting, we train a separate deterministic neural network that serves as an infinite time horizon safety certificate. In particular, we show that the certificate network guarantees the safety of the system over a subset of the BNN weight posterior's support. Our method first computes a safe weight set and then alters the BNN's weight posterior to reject samples outside this set. Moreover, we show how to extend our approach to a safe-exploration reinforcement learning setting, in order to avoid unsafe trajectories during the training of the policy. We evaluate our approach on a series of reinforcement learning benchmarks, including non-Lyapunovian safety specifications. AU - Lechner, Mathias AU - Žikelić, Ðorđe AU - Chatterjee, Krishnendu AU - Henzinger, Thomas A ID - 10667 T2 - 35th Conference on Neural Information Processing Systems TI - Infinite time horizon safety of Bayesian neural networks ER - TY - JOUR AB - For automata, synchronization, the problem of bringing an automaton to a particular state regardless of its initial state, is important. It has several applications in practice and is related to a fifty-year-old conjecture on the length of the shortest synchronizing word. Although using shorter words increases the effectiveness in practice, finding a shortest one (which is not necessarily unique) is NP-hard. For this reason, there exist various heuristics in the literature. However, high-quality heuristics such as SynchroP producing relatively shorter sequences are very expensive and can take hours when the automaton has tens of thousands of states. The SynchroP heuristic has been frequently used as a benchmark to evaluate the performance of the new heuristics. In this work, we first improve the runtime of SynchroP and its variants by using algorithmic techniques. We then focus on adapting SynchroP for many-core architectures, and overall, we obtain more than 1000× speedup on GPUs compared to naive sequential implementation that has been frequently used as a benchmark to evaluate new heuristics in the literature. We also propose two SynchroP variants and evaluate their performance. AU - Sarac, Naci E AU - Altun, Ömer Faruk AU - Atam, Kamil Tolga AU - Karahoda, Sertac AU - Kaya, Kamer AU - Yenigün, Hüsnü ID - 8912 IS - 4 JF - Expert Systems with Applications SN - 09574174 TI - Boosting expensive synchronizing heuristics VL - 167 ER - TY - CONF AB - Formal design of embedded and cyber-physical systems relies on mathematical modeling. In this paper, we consider the model class of hybrid automata whose dynamics are defined by affine differential equations. Given a set of time-series data, we present an algorithmic approach to synthesize a hybrid automaton exhibiting behavior that is close to the data, up to a specified precision, and changes in synchrony with the data. A fundamental problem in our synthesis algorithm is to check membership of a time series in a hybrid automaton. Our solution integrates reachability and optimization techniques for affine dynamical systems to obtain both a sufficient and a necessary condition for membership, combined in a refinement framework. The algorithm processes one time series at a time and hence can be interrupted, provide an intermediate result, and be resumed. We report experimental results demonstrating the applicability of our synthesis approach. AU - Garcia Soto, Miriam AU - Henzinger, Thomas A AU - Schilling, Christian ID - 9200 KW - hybrid automaton KW - membership KW - system identification SN - 9781450383394 T2 - HSCC '21: Proceedings of the 24th International Conference on Hybrid Systems: Computation and Control TI - Synthesis of hybrid automata with affine dynamics from time-series data ER - TY - JOUR AB - A graph game proceeds as follows: two players move a token through a graph to produce a finite or infinite path, which determines the payoff of the game. We study bidding games in which in each turn, an auction determines which player moves the token. Bidding games were largely studied in combination with two variants of first-price auctions called “Richman” and “poorman” bidding. We study taxman bidding, which span the spectrum between the two. The game is parameterized by a constant : portion τ of the winning bid is paid to the other player, and portion to the bank. While finite-duration (reachability) taxman games have been studied before, we present, for the first time, results on infinite-duration taxman games: we unify, generalize, and simplify previous equivalences between bidding games and a class of stochastic games called random-turn games. AU - Avni, Guy AU - Henzinger, Thomas A AU - Žikelić, Đorđe ID - 9239 IS - 8 JF - Journal of Computer and System Sciences SN - 0022-0000 TI - Bidding mechanisms in graph games VL - 119 ER - TY - CONF AB - In runtime verification, a monitor watches a trace of a system and, if possible, decides after observing each finite prefix whether or not the unknown infinite trace satisfies a given specification. We generalize the theory of runtime verification to monitors that attempt to estimate numerical values of quantitative trace properties (instead of attempting to conclude boolean values of trace specifications), such as maximal or average response time along a trace. Quantitative monitors are approximate: with every finite prefix, they can improve their estimate of the infinite trace's unknown property value. Consequently, quantitative monitors can be compared with regard to a precision-cost trade-off: better approximations of the property value require more monitor resources, such as states (in the case of finite-state monitors) or registers, and additional resources yield better approximations. We introduce a formal framework for quantitative and approximate monitoring, show how it conservatively generalizes the classical boolean setting for monitoring, and give several precision-cost trade-offs for monitors. For example, we prove that there are quantitative properties for which every additional register improves monitoring precision. AU - Henzinger, Thomas A AU - Sarac, Naci E ID - 9356 T2 - Proceedings of the 36th Annual ACM/IEEE Symposium on Logic in Computer Science TI - Quantitative and approximate monitoring ER - TY - JOUR AB - Gene expression is regulated by the set of transcription factors (TFs) that bind to the promoter. The ensuing regulating function is often represented as a combinational logic circuit, where output (gene expression) is determined by current input values (promoter bound TFs) only. However, the simultaneous arrival of TFs is a strong assumption, since transcription and translation of genes introduce intrinsic time delays and there is no global synchronisation among the arrival times of different molecular species at their targets. We present an experimentally implementable genetic circuit with two inputs and one output, which in the presence of small delays in input arrival, exhibits qualitatively distinct population-level phenotypes, over timescales that are longer than typical cell doubling times. From a dynamical systems point of view, these phenotypes represent long-lived transients: although they converge to the same value eventually, they do so after a very long time span. The key feature of this toy model genetic circuit is that, despite having only two inputs and one output, it is regulated by twenty-three distinct DNA-TF configurations, two of which are more stable than others (DNA looped states), one promoting and another blocking the expression of the output gene. Small delays in input arrival time result in a majority of cells in the population quickly reaching the stable state associated with the first input, while exiting of this stable state occurs at a slow timescale. In order to mechanistically model the behaviour of this genetic circuit, we used a rule-based modelling language, and implemented a grid-search to find parameter combinations giving rise to long-lived transients. Our analysis shows that in the absence of feedback, there exist path-dependent gene regulatory mechanisms based on the long timescale of transients. The behaviour of this toy model circuit suggests that gene regulatory networks can exploit event timing to create phenotypes, and it opens the possibility that they could use event timing to memorise events, without regulatory feedback. The model reveals the importance of (i) mechanistically modelling the transitions between the different DNA-TF states, and (ii) employing transient analysis thereof. AU - Petrov, Tatjana AU - Igler, Claudia AU - Sezgin, Ali AU - Henzinger, Thomas A AU - Guet, Calin C ID - 9647 JF - Theoretical Computer Science SN - 0304-3975 TI - Long lived transients in gene regulation VL - 893 ER - TY - CONF AB - We argue that the time is ripe to investigate differential monitoring, in which the specification of a program's behavior is implicitly given by a second program implementing the same informal specification. Similar ideas have been proposed before, and are currently implemented in restricted form for testing and specialized run-time analyses, aspects of which we combine. We discuss the challenges of implementing differential monitoring as a general-purpose, black-box run-time monitoring framework, and present promising results of a preliminary implementation, showing low monitoring overheads for diverse programs. AU - Mühlböck, Fabian AU - Henzinger, Thomas A ID - 10108 KW - run-time verification KW - software engineering KW - implicit specification SN - 0302-9743 T2 - International Conference on Runtime Verification TI - Differential monitoring VL - 12974 ER - TY - GEN AB - We argue that the time is ripe to investigate differential monitoring, in which the specification of a program's behavior is implicitly given by a second program implementing the same informal specification. Similar ideas have been proposed before, and are currently implemented in restricted form for testing and specialized run-time analyses, aspects of which we combine. We discuss the challenges of implementing differential monitoring as a general-purpose, black-box run-time monitoring framework, and present promising results of a preliminary implementation, showing low monitoring overheads for diverse programs. AU - Mühlböck, Fabian AU - Henzinger, Thomas A ID - 9946 KW - run-time verification KW - software engineering KW - implicit specification SN - 2664-1690 TI - Differential monitoring ER - TY - JOUR AB - While convolutional neural networks (CNNs) have found wide adoption as state-of-the-art models for image-related tasks, their predictions are often highly sensitive to small input perturbations, which the human vision is robust against. This paper presents Perturber, a web-based application that allows users to instantaneously explore how CNN activations and predictions evolve when a 3D input scene is interactively perturbed. Perturber offers a large variety of scene modifications, such as camera controls, lighting and shading effects, background modifications, object morphing, as well as adversarial attacks, to facilitate the discovery of potential vulnerabilities. Fine-tuned model versions can be directly compared for qualitative evaluation of their robustness. Case studies with machine learning experts have shown that Perturber helps users to quickly generate hypotheses about model vulnerabilities and to qualitatively compare model behavior. Using quantitative analyses, we could replicate users’ insights with other CNN architectures and input images, yielding new insights about the vulnerability of adversarially trained models. AU - Sietzen, Stefan AU - Lechner, Mathias AU - Borowski, Judy AU - Hasani, Ramin AU - Waldner, Manuela ID - 10404 IS - 7 JF - Computer Graphics Forum SN - 0167-7055 TI - Interactive analysis of CNN robustness VL - 40 ER - TY - JOUR AB - In two-player games on graphs, the players move a token through a graph to produce an infinite path, which determines the winner of the game. Such games are central in formal methods since they model the interaction between a non-terminating system and its environment. In bidding games the players bid for the right to move the token: in each round, the players simultaneously submit bids, and the higher bidder moves the token and pays the other player. Bidding games are known to have a clean and elegant mathematical structure that relies on the ability of the players to submit arbitrarily small bids. Many applications, however, require a fixed granularity for the bids, which can represent, for example, the monetary value expressed in cents. We study, for the first time, the combination of discrete-bidding and infinite-duration games. Our most important result proves that these games form a large determined subclass of concurrent games, where determinacy is the strong property that there always exists exactly one player who can guarantee winning the game. In particular, we show that, in contrast to non-discrete bidding games, the mechanism with which tied bids are resolved plays an important role in discrete-bidding games. We study several natural tie-breaking mechanisms and show that, while some do not admit determinacy, most natural mechanisms imply determinacy for every pair of initial budgets. AU - Aghajohari, Milad AU - Avni, Guy AU - Henzinger, Thomas A ID - 10674 IS - 1 JF - Logical Methods in Computer Science KW - computer science KW - computer science and game theory KW - logic in computer science TI - Determinacy in discrete-bidding infinite-duration games VL - 17 ER - TY - CONF AB - Adversarial training is an effective method to train deep learning models that are resilient to norm-bounded perturbations, with the cost of nominal performance drop. While adversarial training appears to enhance the robustness and safety of a deep model deployed in open-world decision-critical applications, counterintuitively, it induces undesired behaviors in robot learning settings. In this paper, we show theoretically and experimentally that neural controllers obtained via adversarial training are subjected to three types of defects, namely transient, systematic, and conditional errors. We first generalize adversarial training to a safety-domain optimization scheme allowing for more generic specifications. We then prove that such a learning process tends to cause certain error profiles. We support our theoretical results by a thorough experimental safety analysis in a robot-learning task. Our results suggest that adversarial training is not yet ready for robot learning. AU - Lechner, Mathias AU - Hasani, Ramin AU - Grosu, Radu AU - Rus, Daniela AU - Henzinger, Thomas A ID - 10666 SN - 1050-4729 T2 - 2021 IEEE International Conference on Robotics and Automation TI - Adversarial training is not ready for robot learning ER - TY - CONF AB - 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. The typical approach is to detect inputs from novel classes and retrain the classifier on an augmented dataset. However, not only the classifier but also the detection mechanism needs to adapt in order to distinguish between newly learned and yet unknown input classes. To address this challenge, we introduce an algorithmic framework for active monitoring of a neural network. A monitor wrapped in our framework operates in parallel with the neural network and interacts with a human user via a series of interpretable labeling queries for incremental adaptation. In addition, we propose an adaptive quantitative monitor to improve precision. An experimental evaluation on a diverse set of benchmarks with varying numbers of classes confirms the benefits of our active monitoring framework in dynamic scenarios. AU - Lukina, Anna AU - Schilling, Christian AU - Henzinger, Thomas A ID - 10206 KW - monitoring KW - neural networks KW - novelty detection SN - 0302-9743 T2 - 21st International Conference on Runtime Verification TI - Into the unknown: active monitoring of neural networks VL - 12974 ER - TY - CONF AB - We propose a neural information processing system obtained by re-purposing the function of a biological neural circuit model to govern simulated and real-world control tasks. Inspired by the structure of the nervous system of the soil-worm, C. elegans, we introduce ordinary neural circuits (ONCs), defined as the model of biological neural circuits reparameterized for the control of alternative tasks. We first demonstrate that ONCs realize networks with higher maximum flow compared to arbitrary wired networks. We then learn instances of ONCs to control a series of robotic tasks, including the autonomous parking of a real-world rover robot. For reconfiguration of the purpose of the neural circuit, we adopt a search-based optimization algorithm. Ordinary neural circuits perform on par and, in some cases, significantly surpass the performance of contemporary deep learning models. ONC networks are compact, 77% sparser than their counterpart neural controllers, and their neural dynamics are fully interpretable at the cell-level. AU - Hasani, Ramin AU - Lechner, Mathias AU - Amini, Alexander AU - Rus, Daniela AU - Grosu, Radu ID - 10673 SN - 2640-3498 T2 - Proceedings of the 37th International Conference on Machine Learning TI - A natural lottery ticket winner: Reinforcement learning with ordinary neural circuits ER - TY - CONF AB - The monitoring of event frequencies can be used to recognize behavioral anomalies, to identify trends, and to deduce or discard hypotheses about the underlying system. For example, the performance of a web server may be monitored based on the ratio of the total count of requests from the least and most active clients. Exact frequency monitoring, however, can be prohibitively expensive; in the above example it would require as many counters as there are clients. In this paper, we propose the efficient probabilistic monitoring of common frequency properties, including the mode (i.e., the most common event) and the median of an event sequence. We define a logic to express composite frequency properties as a combination of atomic frequency properties. Our main contribution is an algorithm that, under suitable probabilistic assumptions, can be used to monitor these important frequency properties with four counters, independent of the number of different events. Our algorithm samples longer and longer subwords of an infinite event sequence. We prove the almost-sure convergence of our algorithm by generalizing ergodic theory from increasing-length prefixes to increasing-length subwords of an infinite sequence. A similar algorithm could be used to learn a connected Markov chain of a given structure from observing its outputs, to arbitrary precision, for a given confidence. AU - Ferrere, Thomas AU - Henzinger, Thomas A AU - Kragl, Bernhard ID - 7348 SN - 1868-8969 T2 - 28th EACSL Annual Conference on Computer Science Logic TI - Monitoring event frequencies VL - 152 ER - TY - CONF AB - We present the results of the ARCH 2020 friendly competition for formal verification of continuous and hybrid systems with linear continuous dynamics. In its fourth edition, eight tools have been applied to solve eight different benchmark problems in the category for linear continuous dynamics (in alphabetical order): CORA, C2E2, HyDRA, Hylaa, Hylaa-Continuous, JuliaReach, SpaceEx, and XSpeed. This report is a snapshot of the current landscape of tools and the types of benchmarks they are particularly suited for. Due to the diversity of problems, we are not ranking tools, yet the presented results provide one of the most complete assessments of tools for the safety verification of continuous and hybrid systems with linear continuous dynamics up to this date. AU - Althoff, Matthias AU - Bak, Stanley AU - Bao, Zongnan AU - Forets, Marcelo AU - Frehse, Goran AU - Freire, Daniel AU - Kochdumper, Niklas AU - Li, Yangge AU - Mitra, Sayan AU - Ray, Rajarshi AU - Schilling, Christian AU - Schupp, Stefan AU - Wetzlinger, Mark ID - 8572 T2 - EPiC Series in Computing TI - ARCH-COMP20 Category Report: Continuous and hybrid systems with linear dynamics VL - 74 ER - TY - CONF AB - We present the results of a friendly competition for formal verification of continuous and hybrid systems with nonlinear continuous dynamics. The friendly competition took place as part of the workshop Applied Verification for Continuous and Hybrid Systems (ARCH) in 2020. This year, 6 tools Ariadne, CORA, DynIbex, Flow*, Isabelle/HOL, and JuliaReach (in alphabetic order) participated. These tools are applied to solve reachability analysis problems on six benchmark problems, two of them featuring hybrid dynamics. We do not rank the tools based on the results, but show the current status and discover the potential advantages of different tools. AU - Geretti, Luca AU - Alexandre Dit Sandretto, Julien AU - Althoff, Matthias AU - Benet, Luis AU - Chapoutot, Alexandre AU - Chen, Xin AU - Collins, Pieter AU - Forets, Marcelo AU - Freire, Daniel AU - Immler, Fabian AU - Kochdumper, Niklas AU - Sanders, David AU - Schilling, Christian ID - 8571 T2 - EPiC Series in Computing TI - ARCH-COMP20 Category Report: Continuous and hybrid systems with nonlinear dynamics VL - 74 ER - TY - CONF AB - A vector addition system with states (VASS) consists of a finite set of states and counters. A transition changes the current state to the next state, and every counter is either incremented, or decremented, or left unchanged. A state and value for each counter is a configuration; and a computation is an infinite sequence of configurations with transitions between successive configurations. A probabilistic VASS consists of a VASS along with a probability distribution over the transitions for each state. Qualitative properties such as state and configuration reachability have been widely studied for VASS. In this work we consider multi-dimensional long-run average objectives for VASS and probabilistic VASS. For a counter, the cost of a configuration is the value of the counter; and the long-run average value of a computation for the counter is the long-run average of the costs of the configurations in the computation. The multi-dimensional long-run average problem given a VASS and a threshold value for each counter, asks whether there is a computation such that for each counter the long-run average value for the counter does not exceed the respective threshold. For probabilistic VASS, instead of the existence of a computation, we consider whether the expected long-run average value for each counter does not exceed the respective threshold. Our main results are as follows: we show that the multi-dimensional long-run average problem (a) is NP-complete for integer-valued VASS; (b) is undecidable for natural-valued VASS (i.e., nonnegative counters); and (c) can be solved in polynomial time for probabilistic integer-valued VASS, and probabilistic natural-valued VASS when all computations are non-terminating. AU - Chatterjee, Krishnendu AU - Henzinger, Thomas A AU - Otop, Jan ID - 8600 SN - 18688969 T2 - 31st International Conference on Concurrency Theory TI - Multi-dimensional long-run average problems for vector addition systems with states VL - 171 ER - TY - CONF AB - A graph game is a two-player zero-sum game in which the players move a token throughout a graph to produce an infinite path, which determines the winner or payoff of the game. In bidding games, both players have budgets, and in each turn, we hold an "auction" (bidding) to determine which player moves the token. In this survey, we consider several bidding mechanisms and study their effect on the properties of the game. Specifically, bidding games, and in particular bidding games of infinite duration, have an intriguing equivalence with random-turn games in which in each turn, the player who moves is chosen randomly. We show how minor changes in the bidding mechanism lead to unexpected differences in the equivalence with random-turn games. AU - Avni, Guy AU - Henzinger, Thomas A ID - 8599 SN - 18688969 T2 - 31st International Conference on Concurrency Theory TI - A survey of bidding games on graphs VL - 171 ER - TY - CONF AB - Machine learning and formal methods have complimentary benefits and drawbacks. In this work, we address the controller-design problem with a combination of techniques from both fields. The use of black-box neural networks in deep reinforcement learning (deep RL) poses a challenge for such a combination. Instead of reasoning formally about the output of deep RL, which we call the wizard, we extract from it a decision-tree based model, which we refer to as the magic book. Using the extracted model as an intermediary, we are able to handle problems that are infeasible for either deep RL or formal methods by themselves. First, we suggest, for the first time, a synthesis procedure that is based on a magic book. We synthesize a stand-alone correct-by-design controller that enjoys the favorable performance of RL. Second, we incorporate a magic book in a bounded model checking (BMC) procedure. BMC allows us to find numerous traces of the plant under the control of the wizard, which a user can use to increase the trustworthiness of the wizard and direct further training. AU - Alamdari, Par Alizadeh AU - Avni, Guy AU - Henzinger, Thomas A AU - Lukina, Anna ID - 9040 SN - 9783854480426 T2 - Proceedings of the 20th Conference on Formal Methods in Computer-Aided Design TI - Formal methods with a touch of magic ER - TY - CONF AB - Second-order information, in the form of Hessian- or Inverse-Hessian-vector products, is a fundamental tool for solving optimization problems. Recently, there has been significant interest in utilizing this information in the context of deep neural networks; however, relatively little is known about the quality of existing approximations in this context. Our work examines this question, identifies issues with existing approaches, and proposes a method called WoodFisher to compute a faithful and efficient estimate of the inverse Hessian. Our main application is to neural network compression, where we build on the classic Optimal Brain Damage/Surgeon framework. We demonstrate that WoodFisher significantly outperforms popular state-of-the-art methods for oneshot pruning. Further, even when iterative, gradual pruning is allowed, our method results in a gain in test accuracy over the state-of-the-art approaches, for standard image classification datasets such as ImageNet ILSVRC. We examine how our method can be extended to take into account first-order information, as well as illustrate its ability to automatically set layer-wise pruning thresholds and perform compression in the limited-data regime. The code is available at the following link, https://github.com/IST-DASLab/WoodFisher. AU - Singh, Sidak Pal AU - Alistarh, Dan-Adrian ID - 9632 SN - 10495258 T2 - Advances in Neural Information Processing Systems TI - WoodFisher: Efficient second-order approximation for neural network compression VL - 33 ER - TY - CONF AB - We introduce LRT-NG, a set of techniques and an associated toolset that computes a reachtube (an over-approximation of the set of reachable states over a given time horizon) of a nonlinear dynamical system. LRT-NG significantly advances the state-of-the-art Langrangian Reachability and its associated tool LRT. From a theoretical perspective, LRT-NG is superior to LRT in three ways. First, it uses for the first time an analytically computed metric for the propagated ball which is proven to minimize the ball’s volume. We emphasize that the metric computation is the centerpiece of all bloating-based techniques. Secondly, it computes the next reachset as the intersection of two balls: one based on the Cartesian metric and the other on the new metric. While the two metrics were previously considered opposing approaches, their joint use considerably tightens the reachtubes. Thirdly, it avoids the "wrapping effect" associated with the validated integration of the center of the reachset, by optimally absorbing the interval approximation in the radius of the next ball. From a tool-development perspective, LRT-NG is superior to LRT in two ways. First, it is a standalone tool that no longer relies on CAPD. This required the implementation of the Lohner method and a Runge-Kutta time-propagation method. Secondly, it has an improved interface, allowing the input model and initial conditions to be provided as external input files. Our experiments on a comprehensive set of benchmarks, including two Neural ODEs, demonstrates LRT-NG’s superior performance compared to LRT, CAPD, and Flow*. AU - Gruenbacher, Sophie AU - Cyranka, Jacek AU - Lechner, Mathias AU - Islam, Md Ariful AU - Smolka, Scott A. AU - Grosu, Radu ID - 9103 SN - 07431546 T2 - Proceedings of the 59th IEEE Conference on Decision and Control TI - Lagrangian reachtubes: The next generation VL - 2020 ER - TY - CONF AB - The family of feedback alignment (FA) algorithms aims to provide a more biologically motivated alternative to backpropagation (BP), by substituting the computations that are unrealistic to be implemented in physical brains. While FA algorithms have been shown to work well in practice, there is a lack of rigorous theory proofing their learning capabilities. Here we introduce the first feedback alignment algorithm with provable learning guarantees. In contrast to existing work, we do not require any assumption about the size or depth of the network except that it has a single output neuron, i.e., such as for binary classification tasks. We show that our FA algorithm can deliver its theoretical promises in practice, surpassing the learning performance of existing FA methods and matching backpropagation in binary classification tasks. Finally, we demonstrate the limits of our FA variant when the number of output neurons grows beyond a certain quantity. AU - Lechner, Mathias ID - 10672 T2 - 8th International Conference on Learning Representations TI - Learning representations for binary-classification without backpropagation ER - TY - CONF AB - Quantization converts neural networks into low-bit fixed-point computations which can be carried out by efficient integer-only hardware, and is standard practice for the deployment of neural networks on real-time embedded devices. However, like their real-numbered counterpart, quantized networks are not immune to malicious misclassification caused by adversarial attacks. We investigate how quantization affects a network’s robustness to adversarial attacks, which is a formal verification question. We show that neither robustness nor non-robustness are monotonic with changing the number of bits for the representation and, also, neither are preserved by quantization from a real-numbered network. For this reason, we introduce a verification method for quantized neural networks which, using SMT solving over bit-vectors, accounts for their exact, bit-precise semantics. We built a tool and analyzed the effect of quantization on a classifier for the MNIST dataset. We demonstrate that, compared to our method, existing methods for the analysis of real-numbered networks often derive false conclusions about their quantizations, both when determining robustness and when detecting attacks, and that existing methods for quantized networks often miss attacks. Furthermore, we applied our method beyond robustness, showing how the number of bits in quantization enlarges the gender bias of a predictor for students’ grades. AU - Giacobbe, Mirco AU - Henzinger, Thomas A AU - Lechner, Mathias ID - 7808 SN - 03029743 T2 - International Conference on Tools and Algorithms for the Construction and Analysis of Systems TI - How many bits does it take to quantize your neural network? VL - 12079 ER - TY - JOUR AB - In resource allocation games, selfish players share resources that are needed in order to fulfill their objectives. The cost of using a resource depends on the load on it. In the traditional setting, the players make their choices concurrently and in one-shot. That is, a strategy for a player is a subset of the resources. We introduce and study dynamic resource allocation games. In this setting, the game proceeds in phases. In each phase each player chooses one resource. A scheduler dictates the order in which the players proceed in a phase, possibly scheduling several players to proceed concurrently. The game ends when each player has collected a set of resources that fulfills his objective. The cost for each player then depends on this set as well as on the load on the resources in it – we consider both congestion and cost-sharing games. We argue that the dynamic setting is the suitable setting for many applications in practice. We study the stability of dynamic resource allocation games, where the appropriate notion of stability is that of subgame perfect equilibrium, study the inefficiency incurred due to selfish behavior, and also study problems that are particular to the dynamic setting, like constraints on the order in which resources can be chosen or the problem of finding a scheduler that achieves stability. AU - Avni, Guy AU - Henzinger, Thomas A AU - Kupferman, Orna ID - 6761 JF - Theoretical Computer Science SN - 03043975 TI - Dynamic resource allocation games VL - 807 ER - TY - CONF AB - Neural networks have demonstrated unmatched performance in a range of classification tasks. Despite numerous efforts of the research community, novelty detection remains one of the significant limitations of neural networks. The ability to identify previously unseen inputs as novel is crucial for our understanding of the decisions made by neural networks. At runtime, inputs not falling into any of the categories learned during training cannot be classified correctly by the neural network. Existing approaches treat the neural network as a black box and try to detect novel inputs based on the confidence of the output predictions. However, neural networks are not trained to reduce their confidence for novel inputs, which limits the effectiveness of these approaches. We propose a framework to monitor a neural network by observing the hidden layers. We employ a common abstraction from program analysis - boxes - to identify novel behaviors in the monitored layers, i.e., inputs that cause behaviors outside the box. For each neuron, the boxes range over the values seen in training. The framework is efficient and flexible to achieve a desired trade-off between raising false warnings and detecting novel inputs. We illustrate the performance and the robustness to variability in the unknown classes on popular image-classification benchmarks. AU - Henzinger, Thomas A AU - Lukina, Anna AU - Schilling, Christian ID - 7505 T2 - 24th European Conference on Artificial Intelligence TI - Outside the box: Abstraction-based monitoring of neural networks VL - 325 ER -