This article we present an intelligent and adaptive system for an autonomous motor-glider based on non-monotonic logic. Piloting is an activity with many constraints. It could have conflicting situations and an uncertain environment. When a pilot faces such kind of situations, he enters into reasoning under uncertainty. Because maybe there are rules with incomplete and contradictory informations, e.i. legislation rules, tower control orders, security rules... Despite the fact that decisions have to be taken at all times regarding different objectives. Default logic is used to find fixed points from ambiguous and conflicting informations. The model proposed here embeds a knowledge of the world based on a set of situations, objectives and actions. After computation of plausible solutions, decision-making is expressed thanks to a non-probabilistic model. On the other hand, we also study the property of resilience which allows to absorb and modify to overcome disturbances. The core of this property is based on the interaction of the possible solutions and the connectivity of its states. Including the resilience in our model allows to carry out the convergence of one or several goals. Eventually, we present some preliminary results of the implementation.

In this work, we introduce a new method for searching stable models of logical programs. This method is based on a relatively new semantics that has not been exploited yet. This semantics captures and extends that one of the stable models (Gelfond et al., 1988) and offers a new alternative to implement ASP solvers. The proposed method performs a DPLL enumerative process that is adapted to Answer Set Programming (ASP) framework according to the used semantics. This method has the advantage to use a Horn clause representation having the same size as the input logic program has constant spatial complexity. It avoids the workload induced by the loop management from which suffer most of the ASP solvers based on the Clark completion. Moreover, the enumeration is done on a restricted set of literals called the strong back-door (STB) of the considered logic program. This reduces the algorithm time complexity which is in theory a function of the size of the STB set. We also introduced new inference rules that the method uses to prune its search tree and hence reduces its size in practice. We implemented the proposed method and applied it to enumerate the stable models of some combinatorial problems. The method is compared to other known systems and the obtained results show that our approach is a good alternative for designing ASP solvers.

International audience

Metabolic pathways are seen as high criticalities in our understanding of mechanisms of biological functions. This article focuses on automatic synthesis of the metabolic pathways from factors in the domain of discrete time. An essential component of this approach is using default logic to acquire facts about biological knowledge of intracellular communication. By choosing an adequate representation of biological knowledge, the "reasoning" is able to assign in acquisition of the facts and extract interactions necessary for the synthesis of the metabolic pathways or signaling.

Many research works had been done in order to define a semantics for logic programs. Most of these semantics are iterated fixed point semantics. The main idea is the canonical model approach which is a declarative semantics for logic programs that can be defined by selecting for each program one of its canonical models. The notion of canonical models of a logic program is what it is called the stable models. The stable models of a logic program are the minimal Her brand models of its "reduct" programs. The work that we describe in this paper is theoretical, we introduce a new semantics for logic programs that is different from the known fixed point semantics. In our approach, logic programs are expressed as CNF formulas (sets of clauses) of a propositional logic for which we define a notion of extension. We prove in this semantics, that each consistent CNF formula admits at least an extension and for each given stable model of a logic program there exists an extension of its corresponding CNF formula which logically entails it. On the other hand, we show that some of the extensions do not entail any stable model, in this case, we define a simple condition called a discrimination condition which allows to recognize such extensions. These extensions could be very important, but are not captured by the stable models semantics. Our approach, extends the stable model semantics in this sense. Following the new semantics, we give a full characterization of the stable models of a logic program by means of the extensions of its CNF encoding verifying the simple discrimination condition, and provide a procedure which can be used to compute such extensions from which we deduce the stable models and eventually the extra-stable models of the given logic program.

In the past decades a lot of researches have been conducted to investigate pathways in order to increase the understanding of the mechanisms responsible for diseases and thus to find new treatments. One strategy to achieve this goal would be to inhibit or to trigger a part of the pathway that produces the disease. A strategy would be to first identify the pathway whose activity contributes to the disease and then look for drugs that inhibit or trigger the pathway acting on its internal components and reactions.

In this paper, a new approach for computing strong backdoor sets of Boolean formula in conjunctive normal form (CNF) is proposed. It makes an original use of local search techniques for finding an assignment leading to a largest renamable Horn sub-formula of a given CNF. More precisely, at each step, preference is given to variables such that when assigned to the opposite value lead to the smallest number of remaining non-Horn clauses. Consequently, if no positive or non Horn clauses remain in the formula, our approach answer the satisfiability of the original formula; otherwise, a smallest non-Horn sub-formula is used to extract the set of variables (strong backdoor) such that when assigned leads to a tractable sub-formula. Branching on the variables of the strong backdoor set leads to significant improvements of Zchaff SAT solver with respect to many real worlds SAT instances

This paper shows how to define nonmonotonic logics from any classical logics ℒ and any set X of formulas of L. In this context, the nonmonotonic inference relation ⊢X is defined by A ⊢XB if every classical theorem of A ∪ B which is in X is a theorem of A. The properties of the relation ⊢ X are studied. We show, in particular, that the elementary properties (supraclassicity, or, left logical equivalence, cut, etc.) are verified for any X. Moreover, we prove that cumulativity is verified if the set of formulas of the language, which are not in X, is deductively closed. Then we prove a representation theorem, i.e. in the finite case every preferential nonmonotonic logic is an X ‐logic. We also study a particular form of the set X for general propositional circumscription.

Hypothesis theory for nonmonotonic reasoning expresses notions of hypotheses and of known information. In this paper, we define these notions in the framework of a new modal system with two modalities, one for exprressing known information and the ot

Revision can be seen as any operation which turns a cognitive state CSt into a subsequent cognitive state CSt'. Two kinds of change can be considered: in the “belief change” case, the cognitive states represent beliefs on a world; they are revised in response to the getting of new information about a static world. In the “world change” case, the cognitive states represent known facts on a real world; they are revised in response to change in this dynamic world. We focus in the following on world change case and propose a way to keep up to date with a dynamic world. Reasoning about change requires predicting how the world will change along time. In absence of a predictive model of evolution, the commonsense law of inertia has been currently used and justifies the minimal change approach to the frame problem. We propose here to use an explicit transition model, which will be used as a predictive evolution model. Dean and Kanazawa propose to use a probabilistic model of persistence and causation. We propose in this paper to use a symbolic model of transition by directly encoding expectations. In the first two sections, we describe the formalism that we propose to explicitly encode the transition model and its axiomatisation. We give then a formal definition of the revision operation using a transition model and discuss what can be a contraction operation in the context of world change. an illustrative example is presented and in the last section, our approach is compared to other related works. © 1994 John Wiley & Sons, Inc.