Luca Pulina

A self-adaptive multi-engine solver for quantified Boolean formulas

In this paper we study the problem of engineering a robust solver for quantified Boolean formulas (QBFs), i.e., a tool that can efficiently solve formulas across different problem domains without the need for domain-specific tuning. The paper presents two main empirical results along this line of research. Our first result is the development of a multi-engine solver, i.e., a tool that selects among its reasoning engines the one which is more likely to yield optimal results. In particular, we show that syntactic QBF features can be correlated to the performances of existing QBF engines across a variety of domains. We also show how a multi-engine solver can be obtained by carefully picking state-of-the-art QBF solvers as basic engines, and by harnessing inductive reasoning techniques to learn engine-selection policies. Our second result is the improvement of our multi-engine solver with the capability of updating the learned policies when they fail to give good predictions. In this way the solver becomes also self-adaptive, i.e., able to adjust its internal models when the usage scenario changes substantially. The rewarding results obtained in our experiments show that our solver AQME—Adaptive QBF Multi-Engine—can be more robust and efficient than state-of-the-art single-engine solvers, even when it is confronted with previously uncharted formulas and competitors.

A multi-engine solver for quantified boolean formulas

In this paper we study the problem of yielding robust performances from current state-of-the-art solvers for quantified Boolean formulas (QBFs). Building on top of existing QBF solvers, we implement a new multi-engine solver which can inductively learn its solver selection strategy. Experimental results confirm that our solver is always more robust than each single engine, that it is stable with respect to various perturbations, and that such results can be partially explained by a handful of features playing a crucial role in our solver.

The seventh QBF solvers evaluation (QBFEVAL’10)

In this paper we report about QBFEVAL’10, the seventh in a series of events established with the aim of assessing the advancements in reasoning about quantified Boolean formulas (QBFs). The paper discusses the results obtained and the experimental setup, from the criteria used to select QBF instances to the evaluation infrastructure. We also discuss the current state-of-the-art in light of past challenges and we envision future research directions that are motivated by the results of QBFEVAL’10.

Evaluating and certifying QBFs: A comparison of state-of-the-art tools

In this paper we compare the performance of all the currently available suites to evaluate and certify QBFs. Our aim is to assess the current state of the art, and also to understand to which extent QBF encodings can be evaluated producing certificates that can be checked in a reliable and efficient way. We conclude that, while the evaluation of some QBFs is still an open challenge, producing and checking certificates for many medium-to-large scale QBFs is feasible with the current technology.

Challenging SMT solvers to verify neural networks

In recent years, Satisfiability Modulo Theory (SMT) solvers are becoming increasingly popular in the Computer Aided Verification and Reasoning community. Used natively or as back-engines, they are accumulating a record of success stories and, as witnessed by the annual SMT competition, their performances and capacity are also increasing steadily. Introduced in previous contributions of ours, a new application domain providing an outstanding challenge for SMT solvers is represented by verification of Multi-Layer Perceptrons (MLPs) a widely-adopted kind of artificial neural network. In this paper we present an extensive evaluation of the current state-of-the-art SMT solvers and assess their potential in the promising domain of MLP verification.

An abstraction-refinement approach to verification of artificial neural networks

A key problem in the adoption of artificial neural networks in safety-related applications is that misbehaviors can be hardly ruled out with traditional analytical or probabilistic techniques In this paper we focus on specific networks known as Multi-Layer Perceptrons (MLPs), and we propose a solution to verify their safety using abstractions to Boolean combinations of linear arithmetic constraints We show that our abstractions are consistent, i.e., whenever the abstract MLP is declared to be safe, the same holds for the concrete one Spurious counterexamples, on the other hand, trigger refinements and can be leveraged to automate the correction of misbehaviors We describe an implementation of our approach based on the HySAT solver, detailing the abstraction-refinement process and the automated correction strategy Finally, we present experimental results confirming the feasibility of our approach on a realistic case study.

Ensuring safety of policies learned by reinforcement: Reaching objects in the presence of obstacles with the iCub

Given a stochastic policy learned by reinforcement, we wish to ensure that it can be deployed on a robot with demonstrably low probability of unsafe behavior. Our case study is about learning to reach target objects positioned close to obstacles, and ensuring a reasonably low collision probability. Learning is carried out in a simulator to avoid physical damage in the trial-and-error phase. Once a policy is learned, we analyze it with probabilistic model checking tools to identify and correct potential unsafe behaviors. The whole process is automated and, in principle, it can be integrated step-by-step with routine task-learning. As our results demonstrate, automated fixing of policies is both feasible and highly effective in bounding the probability of unsafe behaviors.

Treewidth: A Useful Marker of Empirical Hardness in Quantified Boolean Logic Encodings

Theoretical studies show that in some combinatorial problems, there is a close relationship between classes of tractable instances and the treewidth (tw ) of graphs describing their structure. In the case of satisfiability for quantified Boolean formulas (QBFs), tractable classes can be related to a generalization of treewidth, that we call quantified treewidth (tw p ). In this paper we investigate the practical relevance of computing tw p for problem domains encoded as QBFs. We show that an approximation of tw p is a predictor of empirical hardness, and that it is the only parameter among several other candidates which succeeds consistently in being so. We also provide evidence that QBF solvers benefit from a preprocessing phase geared towards reducing tw p , and that such phase is a potential enabler for the solution of hard QBF encodings.

An Ontology for Historical Research Documents

In this paper we present the conceptual layer of stole, our ontology-based digital archive aiming at helping historical researchers to organize data, extract information and derive new knowledge from historical documents.

Safe and effective learning: A case study

In this paper we consider the problem of ensuring that a multi-agent robot control system is both safe and effective in the presence of learning components. Safety, i.e., proving that a potentially dangerous configuration is never reached in the control system, usually competes with effectiveness, i.e., ensuring that tasks are performed at an acceptable level of quality. In particular, we focus on a robot playing the air hockey game against a human opponent, where the robot has to learn how to minimize opponents goals (defense play). This setup is paradigmatic since the robot must see, decide and move fastly, but, at the same time, it must learn and guarantee that the control system is safe throughout the process. We attack this problem using automata-theoretic formalisms and associated verification tools, showing experimentally that our approach can yield safety without heavily compromising effectiveness.