Abstract In this paper, we address the problem of giving names to predicates in logic rules using Large Language Models (LLMs). In the context of Inductive Logic Programming, various rule generation methods produce rules containing unnamed predicates, with Predicate Invention being a key example. This hinders the readability, interpretability, and reusability of the logic theory. Leveraging recent advancements in LLMs development, we explore their ability to process natural language and code to provide semantically meaningful suggestions for giving a name to unnamed predicates. The evaluation of our approach on some hand-crafted logic rules indicates that LLMs hold potential for this task.

Learning from interpretation transition (LFIT) automatically constructs a model of the dynamics of a system from the observation of its state transitions. So far, the systems that LFIT handles are restricted to discrete variables or suppose a discretization of continuous data. However, when working with real data, the discretization choices are critical for the quality of the model learned by LFIT . In this paper, we focus on a method that learns the dynamics of the system directly from continuous time-series data. For this purpose, we propose a modeling of continuous dynamics by logic programs composed of rules whose conditions and conclusions represent continuums of values.

Probabilistic planners have improved recently to the point that they can solve difficult tasks with complex and expressive models. In contrast, learners cannot tackle yet the expressive models that planners do, which forces complex models to be mostly handcrafted. We propose a new learning approach that can learn relational probabilistic models with both action effects and exogenous effects. The proposed learning approach combines a multi-valued variant of inductive logic programming for the generation of candidate models, with an optimization method to select the best set of planning operators to model a problem. We also show how to combine this learner with reinforcement learning algorithms to solve complete problems. Finally, experimental validation is provided that shows improvements over previous work in both simulation and a robotic task. The robotic task involves a dynamic scenario with several agents where a manipulator robot has to clear the tableware on a table. We show that the exogenous effects learned by our approach allowed the robot to clear the table in a more efficient way.

Delayed effects are important in modeling biological systems, and timed Boolean networks have been proposed for such a framework. Yet it is not an easy task to design such Boolean models with delays precisely. Recently, an attempt to learn timed Boolean networks has been made in Ribeiro et al 2015 in the framework of learning state transition rules from time-series data. However, this approach still has two limitations: (1) The maximum delay has to be given as input to the algorithm, (2) The possible value of each state is assumed to be Boolean, i.e., twovalued. In this paper, we extend the previous learning mechanism to overcome these limitations. We propose an algorithm to learn multi-valued biological models with delayed influence by automatically tuning the delay. The delay is determined so as to minimally explain the necessary influences. The merits of our approach is then verified on benchmarks coming from the DREAM4 challenge.

A Constraint Satisfaction Problem (CSP) is a fundamental problem that can formalize various applications related to Artificial Intelligence problems. A Weighted Constraint Satisfaction Problem (WCSP) is a CSP where constraints can be violated, and the aim of this problem is to find an assignment that minimizes the sum of weights of the violated constraints. Most researches have focused on developing algorithms for solv- ing static mono-objective problems. However, many real world satisfaction/optimization problems involve multiple criteria that should be considered separately and satisfied/optimized simultaneously. Additionally, they are often dynamic, i.e., the problem changes at runtime. In this paper, we introduce a Multi-Objective WCSP (MO-WCSP) and develop a novel MO-WCSP algorithm called Multi-Objective Branch and Bound (MO-BnB), which is based on a new solution criterion called (l, s)-Pareto solution. Furthermore, we first for- malize a Dynamic MO-WCSP (DMO-WCSP). As an initial step forward developing an algorithm for solving a DMO-WCSP, we focus on the change of weights of constraints and develop the first algorithm called Dynamic Multi-Objective Branch and Bound (DMO-BnB) for solving a DMO-WCSPs, which is based on MO-BnB. Finally, we provide the complexity of our algorithm and evaluate DMO-BnB with different problem settings.