A core task in process mining is process discovery which aims to learn an accurate process model from event log data. In this paper, we propose to use (block-) structured programs directly as target process models so as to establish connections to the field of program synthesis and facilitate the translation from abstract process models to executable processes, e.g., for robotic process automation. Furthermore, we develop a novel bottom-up agglomerative approach to the discovery of such structured program process models. In comparison with the popular top-down recursive inductive miner, our proposed agglomerative miner enjoys the similar theoretical guarantee to produce sound process models (without deadlocks and other anomalies) while exhibiting some advantages like avoiding silent activities and accommodating duplicate activities. The proposed algorithm works by iteratively applying a few graph rewriting rules to the directly-follows-graph of activities. For real-world (sparse) directly-follows-graphs, the algorithm has quadratic computational complexity with respect to the number of distinct activities. To our knowledge, this is the first process discovery algorithm that is made for the purpose of program synthesis. Experiments on the BPI-Challenge 2020 dataset and the Karel programming dataset have demonstrated that our proposed algorithm can outperform the inductive miner not only according to the traditional process discovery metrics but also in terms of the effectiveness in finding out the true underlying structured program from a small number of its execution traces.

We examine whether self-supervised language modeling applied to mathematical formulas enables logical reasoning. We suggest several logical reasoning tasks that can be used to evaluate language models trained on formal mathematical statements, such as type inference, suggesting missing assumptions and completing equalities. To train language models for formal mathematics, we propose a novel skip-tree task. We find that models trained on the skip-tree task show surprisingly strong mathematical reasoning abilities, and outperform models trained on standard skip-sequence tasks. We also analyze the models' ability to formulate new conjectures by measuring how often the predictions are provable and useful in other proofs.

This paper continues the line of research aimed at investigating the relationship between logic programs and first-order theories. We extend the definition of program completion to programs with input and output in a subset of the input language of the ASP grounder gringo, study the relationship between stable models and completion in this context, and describe preliminary experiments with the use of two software tools, anthem and vampire, for verifying the correctness of programs with input and output. Proofs of theorems are based on a lemma that relates the semantics of programs studied in this paper to stable models of first-order formulas. This paper is under consideration for acceptance in Theory and Practice of Logic Programming.

Action selection policies (ASPs), used to compose low-level robot skills into complex high-level tasks are commonly represented as neural networks (NNs) in the state of the art. Such a paradigm, while very effective, suffers from a few key problems: 1) NNs are opaque to the user and hence not amenable to verification, 2) they require significant amounts of training data, and 3) they are hard to repair when the domain changes. We present two key insights about ASPs for robotics. First, ASPs need to reason about physically meaningful quantities derived from the state of the world, and second, there exists a layered structure for composing these policies. Leveraging these insights, we introduce layered dimension-informed program synthesis (LDIPS) - by reasoning about the physical dimensions of state variables, and dimensional constraints on operators, LDIPS directly synthesizes ASPs in a human-interpretable domain-specific language that is amenable to program repair. We present empirical results to demonstrate that LDIPS 1) can synthesize effective ASPs for robot soccer and autonomous driving domains, 2) requires two orders of magnitude fewer training examples than a comparable NN representation, and 3) can repair the synthesized ASPs with only a small number of corrections when transferring from simulation to real robots.

Recent trends in AI verification and Explainable AI have raised the question of whether AI planning techniques can be verified. In this paper, we present a novel resource logic, the Proof Carrying Plans (PCP) logic that can be used to verify plans produced by AI planners. The PCP logic takes inspiration from existing resource logics (such as Linear logic and Separation logic) as well as Hoare logic when it comes to modelling states and resource-aware plan execution. It also capitalises on the Curry-Howard approach to logics, in its treatment of plans as functions and plan pre- and post-conditions as types. This paper presents two main results. From the theoretical perspective, we show that the PCP logic is sound relative to the standard possible world semantics used in AI planning. From the practical perspective, we present a complete Agda formalisation of the PCP logic and of its soundness proof. Moreover, we showcase the Curry-Howard, or functional, value of this implementation by supplementing it with the library that parses AI plans into Agda's proofs automatically. We provide evaluation of this library and the resulting Agda functions.

Weighted Logic is a powerful tool for the specification of calculations over semirings that depend on qualitative information. Using a novel combination of Weighted Logic and Here-and-There (HT) Logic, in which this dependence is based on intuitionistic grounds, we introduce Answer Set Programming with Algebraic Constraints (ASP(A C)), where rules may contain constraints that compare semiring values to weighted formula evaluations. Such constraints provide streamlined access to a manifold of constructs available in ASP, like aggregates, choice constraints, and arithmetic operators. They extend some of them and provide a generic framework for defining programs with algebraic computation, which can be fruitfully used e.g. for provenance semantics of datalog programs. While undecidable in general, expressive fragments of ASP(A C) can be exploited for effective problem solving in a rich framework. This work is under consideration for acceptance in Theory and Practice of Logic Programming.

We propose a novel formal framework (called 3D-nCDC-ASP) to represent and reason about cardinal directions between extended objects in 3-dimensional (3D) space, using Answer Set Programming (ASP). 3D-nCDC-ASP extends Cardinal Directional Calculus (CDC) with a new type of default constraints, and nCDC-ASP to 3D. 3D-nCDC-ASP provides a flexible platform offering different types of reasoning: Nonmonotonic reasoning with defaults, checking consistency of a set of constraints on 3D cardinal directions between objects, explaining inconsistencies, and inferring missing CDC relations. We prove the soundness of 3D-nCDC-ASP, and illustrate its usefulness with applications. This paper is under consideration for acceptance in TPLP.

This paper studies novel subgoaling relaxations for automated planning with propositional and numeric state variables. Subgoaling relaxations address one source of complexity of the planning problem: the requirement to satisfy conditions simultaneously. The core idea is to relax this requirement by recursively decomposing conditions into atomic subgoals that are considered in isolation. Such relaxations are typically used for pruning, or as the basis for computing admissible or inadmissible heuristic estimates to guide optimal or satisificing heuristic search planners. In the last decade or so, the subgoaling principle has underpinned the design of an abundance of relaxation-based heuristics whose formulations have greatly extended the reach of classical planning. This paper extends subgoaling relaxations to support numeric state variables and numeric conditions. We provide both theoretical and practical results, with the aim of reaching a good trade-off between accuracy and computation costs within a heuristic state-space search planner. Our experimental results validate the theoretical assumptions, and indicate that subgoaling substantially improves on the state of the art in optimal and satisficing numeric planning via forward state-space search.

We focus on the problem of inducing logic programs that explain models learned by the support vector machine (SVM) algorithm. The top-down sequential covering inductive logic programming (ILP) algorithms (e.g., FOIL) apply hill-climbing search using heuristics from information theory. A major issue with this class of algorithms is getting stuck in a local optimum. In our new approach, however, the data-dependent hill-climbing search is replaced with a model-dependent search where a globally optimal SVM model is trained first, then the algorithm looks into support vectors as the most influential data points in the model, and induces a clause that would cover the support vector and points that are most similar to that support vector. Instead of defining a fixed hypothesis search space, our algorithm makes use of SHAP, an example-specific interpreter in explainable AI, to determine a relevant set of features. This approach yields an algorithm that captures SVM model's underlying logic and outperforms %GG: the FOIL algorithm --> other ILP algorithms other ILP algorithms in terms of the number of induced clauses and classification evaluation metrics. This paper is under consideration for publication in the journal of "Theory and practice of logic programming".

If you read a description of a static analysis in a paper, what might you find? There'll be some cute model of a language. Maybe some inference rules describing the analysis itself, but those rules probably rely on a variety of helper functions. These days, the analysis likely involves some logical reasoning: about the terms in the language, the branches conditionals might take, and so on. What makes a language good for implementing such an analysis? You'd want a variety of features: Aaron Bembenek, Steve Chong, and I have developed a design that hits the sweet spot of those four points: given Datalog as a core, you add constructors, pure ML, and a type-safe interface to SMT.