The increasing deployment of AI (artificial intelligence) on edge devices presents major challenges due to strict constraints on computation, memory, energy, and latency. Effective Edge AI systems thus require multi-objective optimization that balances accuracy, hardware efficiency, and reliability. The Horizon Twinning project AIDA4Edge tackles these challenges by developing methods for efficient and reliable AI on resource-constrained platforms. This paper presents key approaches explored within the project, including neural network quantization, hardware-aware neural architecture search, dynamic neural networks, and self-adaptive resilient AI architectures. Finally, these strategies are placed within a broader, biologically inspired paradigm, highlighting neuromorphic computing as a natural continuation of Edge AI efforts toward highly efficient and resilient intelligent systems.

The growing demand for deployment of Artificial Intelligence (AI) on resource-constrained edge devices has motivated extensive research on the design of efficient edge-compatible AI hardware accelerators. One of the most promising solutions are the self-adaptive AI accelerators, capable of optimizing in real time their performance and energy consumption according to application requirements. This work introduces the EU-funded project Twinning for Excellence in Adaptive Edge Artificial Intelligence (AIDA4Edge), aimed to advance the state-of-the-art in the design of adaptive neural network accelerators for edge applications. The main goal is to develop a novel hybrid self-adaptive neural network architecture combining spiking and artificial neural networks, and supporting runtime adaptation of network functionality, precision and reliability. Furthermore, we aim to enhance the neural network training by incorporating hardware and quantization constraints in an automated tuning engine.

Hybrid Knowledge Bases (HKBs) successfully integrate Logic Programming (LP) and Description Logics (DL) under the Minimal Knowledge with Negation as Failure semantics. Both world closure assumptions (open and closed) can be used in the same HKB, a feature required in many domains, such as the legal and health-care ones. In previous work, we proposed (function-free) Probabilistic HKBs, whose semantics applied Sato's distribution semantics approach to the well-founded HKB semantics proposed by Knorr et al. and Lyu and You. This semantics relied on the fact that the grounding of a function-free Probabilistic HKB (PHKB) is finite. In this article, we extend the PHKB language to allow function symbols, obtaining PHKBFS. Because the grounding of a PHKBFS can be infinite, we propose a novel semantics which does not require the PHKBFS's grounding to be finite. We show that the proposed semantics extends the previously proposed semantics and that, for a large class of PHKBFS, every query can be assigned a probability.

The necessity to manage inconsistency in Description Logics Knowledge Bases (KBs) has come to the fore with the increasing importance gained by the Semantic Web, where information comes from different sources that constantly change their content and may contain contradictory descriptions when considered either alone or together. Classical reasoning algorithms do not handle inconsistent KBs, forcing the debugging of the KB in order to remove the inconsistency. In this paper, we exploit an existing probabilistic semantics called DISPONTE to overcome this problem and allow queries also in case of inconsistent KBs. We implemented our approach in the reasoners TRILL and BUNDLE and empirically tested the validity of our proposal. Moreover, we formally compare the presented approach to that of the repair semantics, one of the most established semantics when considering DL reasoning tasks.

Abstract Probabilistic Logic Programming combines uncertainty and logic-based languages. Liftable Probabilistic Logic Programs have been recently proposed to perform inference in a lifted way. LIFTCOVER is an algorithm used to perform parameter and structure learning of liftable probabilistic logic programs. In particular, it performs parameter learning via Expectation Maximization and LBFGS. In this paper, we present an updated version of LIFTCOVER, called LIFTCOVER+, in which regularization was added to improve the quality of the solutions and LBFGS was replaced by gradient descent. We tested LIFTCOVER+ on the same 12 datasets on which LIFTCOVER was tested and compared the performances in terms of AUC-ROC, AUC-PR, and execution times. Results show that in most cases Expectation Maximization with regularization improves the quality of the solutions.

Abstract The aim of the Semantic Web is making information and resources from the Web automatically processable by machines. Usually, the uncertainty characterizing much of this information is addressed by means of a probabilistic semantics. Following the vision of a “Probabilistic Semantic Web”, a plethora of probabilistic semantics have been proposed: some of them change the syntax and/or the semantics itself of the knowledge representation language, others allow one to annotate axioms of a knowledge base with a probability value. Among the latter, the DISPONTE semantics exploits probabilistic annotations to extend query answering with the capability of returning the probability of a query being true in a domain. In order to promote the adoption of Probabilistic Semantic Web we first developed BUNDLE, a framework that can exploit different underlying (probabilistic and non‐probabilistic) reasoners to perform probabilistic inference under the DISPONTE semantics. In this paper we present a web application for BUNDLE, to show how DISPONTE is easily usable even in already established applications and systems. It allows users to query a DISPONTE knowledge base written or uploaded directly in the application interface by using just a web browser, without the need to install any software on their machine. It is accessible on the web at https://bundle.ml.unife.it/ and also provides some examples for familiarizing with the application. The results of a usability evaluation involving human participants are also reported, showing the relevance and the practical impact of the tool and possible ways for improvement.

Abstract Probabilistic logical models deal effectively with uncertain relations and entities typical of many real world domains. In the field of probabilistic logic programming usually the aim is to learn these kinds of models to predict specific atoms or predicates of the domain, called target atoms/predicates. However, it might also be useful to learn classifiers for interpretations as a whole: to this end, we consider the models produced by the inductive constraint logic system, represented by sets ofintegrity constraints, and we propose a probabilistic version of them. Each integrity constraint is annotated with a probability, and the resulting probabilistic logical constraint model assigns a probability of being positive to interpretations. To learn both the structure and the parameters of such probabilistic models we propose the system PASCAL for “probabilistic inductive constraint logic”. Parameter learning can be performed using gradient descent or L-BFGS. PASCAL has been tested on 11 datasets and compared with a few statistical relational systems and a system that builds relational decision trees (TILDE): we demonstrate that this system achieves better or comparable results in terms of area under the precision–recall and receiver operating characteristic curves, in a comparable execution time.

Abstract Uncertain information is being taken into account in an increasing number of application fields. In the meantime, abduction has been proved a powerful tool for handling hypothetical reasoning and incomplete knowledge. Probabilistic logical models are a suitable framework to handle uncertain information, and in the last decade many probabilistic logical languages have been proposed, as well as inference and learning systems for them. In the realm of Abductive Logic Programming (ALP), a variety of proof procedures have been defined as well. In this paper, we consider a richer logic language, coping with probabilistic abduction with variables. In particular, we consider an ALP program enriched with integrity constraints à la IFF, possibly annotated with a probability value. We first present the overall abductive language and its semantics according to the Distribution Semantics. We then introduce a proof procedure, obtained by extending one previously presented, and prove its soundness and completeness.

Abductive Logic Programming (ALP) has been proven very effective for formalizing societies of agents, commitments and norms, in particular by mapping the most common deontic operators (obligation, prohibition, permission) to abductive expectations. In our previous works, we have shown that ALP is a suitable framework for representing norms. Normative reasoning and query answering were accommodated by the same abductive proof procedure, named 𝒮CIFF. In this work, we introduce a defeasible flavour in this framework, in order to possibly discharge obligations in some scenarios. Abductive expectations can also be qualified as dischargeable, in the new, extended syntax. Both declarative and operational semantics are improved accordingly, and proof of soundness is given under syntax allowedness conditions Moreover, the dischargement itself might be proved invalid, or incoherent with the rules, due to new knowledge provided later on. In such a case, a discharged expectation might be reinstated and hold again after some evidence is given. We extend the notion of dischargement to take into consideration also the reinstatement of expectations. The expressiveness and power of the extended framework, named 𝒮CIFF𝒟, is shown by modeling and reasoning upon a fragment of the Japanese Civil Code. In particular, we consider a case study concerning manifestations of intention and their rescission (Section II of the Japanese Civil Code).

Abstract In Probabilistic Logic Programming (PLP) the most commonly studied inference task is to compute the marginal probability of a query given a program. In this paper, we consider two other important tasks in the PLP setting: the Maximum-A-Posteriori (MAP) inference task, which determines the most likely values for a subset of the random variables given evidence on other variables, and the Most Probable Explanation (MPE) task, the instance of MAP where the query variables are the complement of the evidence variables. We present a novel algorithm, included in the PITA reasoner, which tackles these tasks by representing each problem as a Binary Decision Diagram and applying a dynamic programming procedure on it. We compare our algorithm with the version of ProbLog that admits annotated disjunctions and can perform MAP and MPE inference. Experiments on several synthetic datasets show that PITA outperforms ProbLog in many cases.

Abstract When modeling real-world domains, we have to deal with information that is incomplete or that comes from sources with different trust levels. This motivates the need for managing uncertainty in the Semantic Web. To this purpose, we introduced a probabilistic semantics, named DISPONTE, in order to combine description logics (DLs) with probability theory. The probability of a query can be then computed from the set of its explanations by building a Binary Decision Diagram (BDD). The set of explanations can be found using thetableau algorithm, which has to handle non-determinism. Prolog, with its efficient handling of non-determinism, is suitable for implementing the tableau algorithm. TRILL and TRILLPare systems offering a Prolog implementation of the tableau algorithm. TRILLPbuilds apinpointing formulathat compactly represents the set of explanations and can be directly translated into a BDD. Both reasoners were shown to outperform state-of-the-art DL reasoners. In this paper, we present an improvement of TRILLP, named TORNADO, in which the BDD is directly built during the construction of the tableau, further speeding up the overall inference process. An experimental comparison shows the effectiveness of TORNADO. All systems can be tried online in the TRILL on SWISH web application at http://trill.ml.unife.it/ .

Ontologies form the basis of the Semantic Web. Description Logics (DLs) are often the languages of choice for modeling ontologies. Integration of DLs with rules and rule-based reasoning is crucial in the so-called Semantic Web stack vision - a complete stack of recommendations and languages each ba sed on and/or exploiting the underlying layers - which adds new features to the standards used in theWeb. The growing importance of the integration between DLs and rules is proved by the definition of the profile OWL 2 RL1 and the definition of languages such as RIF2 and SWRL3. Datalog± is an extension of Datalog which can be used for representing lightweight ontologies and expressing some languages of the DL-Lite family, with tractable query answering under certain language restrictions. In particular, it is able to express the DL-Lite version defined in OWL. In this work, we show that Abductive Logic Programming (ALP) can be used to represent Datalog± ontologies, supporting query answering through an abductive proof procedure, and smoothly achieving the integration of ontologies and rule-based reasoning. Often, reasoning with DLs means finding explanations for the truth of queries, that are useful when debugging ontologies and to understand answers given by the reasoning process. We show that reasoning under existential rules can be expressed by ALP languages and we present a solving system, which is experimentally proved to be competitive with DL reasoning systems. In particular, we consider an ALP framework named 𝒮CIFF derived from the IFF abductive framework. Forward and backward reasoning is naturally supported in this ALP framework. The 𝒮CIFF language smoothly supports the integration of rules, expressed in a Logic Programming language, with Datalog± ontologies, mapped into 𝒮CIFF (forward) integrity constraints. The main advantage is that this integration is achieved within a single language, grounded on abduction in computational logic, and able to model existential rules.

We present the web application Tableau Reasoner for descrIption Logics in proLog on SWI-Prolog for SHaring (TRILL on SWISH) which allows the user to write probabilistic description logic (DL) theories and compute the probability of queries with just a web browser. Various probabilistic extensions of DLs have been proposed in the recent past, because uncertainty is a fundamental component of the Semantic Web. We consider probabilistic DL theories following our distribution semantics for probabilistic ontologies (DISPONTE) semantics. Axioms of a DISPONTE knowledge base can be annotated with a probability, and the probability of queries can be computed with inference algorithms. TRILL is a probabilistic reasoner for DISPONTE knowledge base that is implemented in Prolog and exploits its backtracking facilities for handling the non-determinism of the tableau algorithm. TRILL on SWISH is based on SWISH, a recently proposed web framework for logic programming, based on various features and packages of SWI-Prolog (e.g., a web server and a library for creating remote Prolog engines and posing queries to them). TRILL on SWISH also allows users to cooperate in writing a probabilistic DL theory. It is free, open, and accessible on the Web at the url: http://trill.lamping.unife.it; it includes a number of examples that cover a wide range of domains and provide interesting Probabilistic Semantic Web applications. By building a web-based system, we allow users to experiment with probabilistic DLs without the need to install a complex software stack. In this way, we aim to reach out to a wider audience and popularize the Probabilistic Semantic Web. Copyright © 2016 John Wiley & Sons, Ltd.

Probabilistic Logic Programming is a promising formalism for dealing with uncertainty. Learning probabilistic logic programs has been receiving an increasing attention in Inductive Logic Programming: for instancethe system SLIPCOVER learns high quality theories in a variety of domains. HoweverSLIPCOVER is computationally expensivewith a running time of the order of hours. In order to apply SLIPCOVER to Big Data, we present SEMPRE, for “Structure lEarning by MaPREduce”, that scales SLIPCOVER by following a MapReduce strategy, directly implemented with the Message Passing Interface.

Summary We present the web application ‘cplinton SWI‐Prolog for SHaring that allows the user to write (SWISH)' Probabilistic Logic Programs and submit the computation of the probability of queries with a web browser. The application is based on SWISH, a web framework for Logic Programming. SWISH is based on various features and packages of SWI‐Prolog, in particular, its web server and its Pengine library, that allow to create remote Prolog engines and to pose queries to them. In order to develop the web application, we started from the PITA system, which is included incplint, a suite of programs for reasoning over Logic Programs with Annotated Disjunctions, by porting PITA to SWI‐Prolog. Moreover, we modified the PITA library so that it can be executed in a multi‐threading environment. Developing ‘cplinton SWISH’ also required modification of the JavaScript SWISH code that creates and queries Pengines. ‘cplinton SWISH’ includes a number of examples that cover a wide range of domains and provide interesting applications of Probabilistic Logic Programming. By providing a web interface tocplint, we allow users to experiment with Probabilistic Logic Programming without the need to install a system, a procedure that is often complex, error prone, and limited mainly to the Linux platform. In this way, we aim to reach out to a wider audience and popularize Probabilistic Logic Programming. Copyright © 2015 John Wiley & Sons, Ltd.

Abstract Lifted inference has been proposed for various probabilistic logical frameworks in order to compute the probability of queries in a time that depends on the size of the domains of the random variables rather than the number of instances. Even if various authors have underlined its importance for probabilistic logic programming (PLP), lifted inference has been applied up to now only to relational languages outside of logic programming. In this paper we adapt Generalized Counting First Order Variable Elimination (GC-FOVE) to the problem of computing the probability of queries to probabilistic logic programs under the distribution semantics. In particular, we extend the Prolog Factor Language (PFL) to include two new types of factors that are needed for representing ProbLog programs. These factors take into account the existing causal independence relationships among random variables and are managed by the extension to variable elimination proposed by Zhang and Poole for dealing with convergent variables and heterogeneous factors. Two new operators are added to GC-FOVE for treating heterogeneous factors. The resulting algorithm, called LP2for Lifted Probabilistic Logic Programming, has been implemented by modifying the PFL implementation of GC-FOVE and tested on three benchmarks for lifted inference. A comparison with PITA and ProbLog2 shows the potential of the approach.