Abstract Logic-based machine learning aims to learn general, interpretable knowledge in a data-efficient manner. However, labelled data must be specified structured logical form. To address this limitation, we propose neural-symbolic framework, called Feed-Forward Neural-Symbolic Learner (FFNSL) , that integrates logic-based system capable of from noisy examples, with neural networks, order unstructured data. We demonstrate the generality FFNSL on four classification problems, where...

Artificial Intelligence is rapidly enhancing human capability by providing support and guidance on a wide variety of tasks. However, one the main challenges for autonomous systems effectively managing decisions interactions between multiple entities in dynamic environment. Policies are frequently used cyber-physical to define target goals constraints, such as maximising security whilst preventing communication unauthorised systems. In this paper we introduce an approach learning high-level...

One of the ultimate goals Artificial Intelligence is to assist humans in complex decision making. A promising direction for achieving this goal Neuro-Symbolic AI, which aims combine interpretability symbolic techniques with ability deep learning learn from raw data. However, most current approaches require manually engineered knowledge, and where end-to-end training considered, such are either restricted definite programs, or binary neural networks. In paper, we introduce Inductive Learner...

The way we travel is changing rapidly and Cooperative Intelligent Transportation Systems (C-ITSs) are at the forefront of this evolution. However, adoption C-ITSs introduces new risks challenges, making cybersecurity a top priority for ensuring safety reliability. Building on premise, paper an envisaged Cybersecurity Centre Excellence (CSCE) designed to bolster researching, testing, evaluating C-ITSs. We explore design, functionality, challenges CSCE's testing facilities, outlining...

Concept Bottleneck Models (CBMs) are considered inherently interpretable because they first predict a set of human-defined concepts before using these to the output downstream task. For inherent interpretability be fully realised, and ensure trust in model's output, we need guarantee predicted based on semantically mapped input features. example, one might expect pixels representing broken bone an image used for prediction fracture. However, current literature indicates this is not case, as...

Neuro-Symbolic AI (NeSy) holds promise to ensure the safe deployment of systems, as interpretable symbolic techniques provide formal behaviour guarantees. The challenge is how effectively integrate neural and computation, enable learning reasoning from raw data. Existing pipelines that train components sequentially require extensive labelling, whereas end-to-end approaches are limited in terms scalability, due combinatorial explosion symbol grounding problem. In this paper, we leverage...

Concept Bottleneck Models (CBMs) first map raw input(s) to a vector of human-defined concepts, before using this predict final classification. We might therefore expect CBMs capable predicting concepts based on distinct regions an input. In doing so, would support human interpretation when generating explanations the model's outputs visualise input features corresponding concepts. The contribution paper is threefold: Firstly, we expand existing literature by looking at relevance both from...

One of the ultimate goals Artificial Intelligence is to assist humans in complex decision making. A promising direction for achieving this goal Neuro-Symbolic AI, which aims combine interpretability symbolic techniques with ability deep learning learn from raw data. However, most current approaches require manually engineered knowledge, and where end-to-end training considered, such are either restricted definite programs, or binary neural networks. In paper, we introduce Inductive Learner...

Given the rapid increase in urbanisation and a change mobility patterns of humans, traffic on our road networks is expanding at an exponential rate. Governments across globe are investing vast amount money to cater for this ever increasing demand. This, however adds cognitive burden driver as moving parts network also increased. Motivated by observation, paper, we hypothesise that advances technology-be it connected cars or smart infrastructures- could play key role effectively efficiently...

Policy systems are critical for managing missions and collaborative activities carried out by coalitions involving different organizations. Conventional policy-based management approaches not suitable next-generation that will involve only humans, but also autonomous computing devices systems. It is those parties be able to generate customize policies based on contexts activities. This paper introduces a novel approach the autonomic generation of parties. The framework combines context free...

This paper investigates neural-symbolic policy learning for information fusion in distributed human-agent operations. The architecture integrates a pre-trained neural network feature extraction, with state-of-the-art symbolic Inductive Logic Programming (ILP) system to learn policies, expressed as set of logical rules. We firstly outline the challenge within military environment, by investigating accuracy and confidence predictions given data outside training distribution. Secondly, we...

Autonomous systems are expected to have a major impact in future coalition operations. These enabled by variety of Artificial Intelligence (AI) learning algorithms that contextualize and adapt varying, possibly unforeseen situations assist humans achieving complex tasks. Moreover, these will be required operate dynamic challenging environments impose certain constraints such as task formation collaboration, ad-hoc resource availability, rapidly changing environmental conditions the...

To facilitate information sharing between systems and devices in a distributed environment, unstructured data from various sensors must be analysed accordingly. Recent work has developed the notion of context-dependant generative policy framework capable learning models strings text-based tabular format. However, it is vital that contextual can alongside data, potentially at edge network to enable applied more complex tasks. This paper performs deep-dive into field neuralsymbolic machine...

Inductive Logic Programming (ILP) systems learn generalised, interpretable rules in a data-efficient manner utilising existing background knowledge. However, current ILP require training examples to be specified structured logical format. Neural networks from unstructured data, although their learned models may difficult interpret and are vulnerable data perturbations at run-time. This paper introduces hybrid neural-symbolic learning framework, called NSL, that learns labelled data. NSL...

Discovering new materials is essential to solve challenges in climate change, sustainability and healthcare. A typical task discovery search for a material database which maximises the value of function. That function often expensive evaluate, can rely upon simulation or an experiment. Here, we introduce SyMDis, sample efficient optimisation method based on symbolic learning, that discovers near-optimal large database. SyMDis performs comparably state-of-the-art optimiser, whilst learning...

The way we travel is changing rapidly, and Cooperative Intelligent Transportation Systems (C-ITSs) are at the forefront of this evolution. However, adoption C-ITSs introduces new risks challenges, making cybersecurity a top priority for ensuring safety reliability. Building on premise, paper presents an envisaged Cybersecurity Centre Excellence (CSCE) designed to bolster research, testing, evaluation C-ITSs. We explore design, functionality, challenges CSCE's testing facilities, outlining...

Generative Policy Models (GPMs) have been proposed as a method for future autonomous decision making in distributed, collaborative environment. To learn GPM, previous policy examples that contain features and the corresponding decisions are used. Recently, GPMs constructed using both symbolic statistical learning algorithms. In either case, goal of process is to create model across wide range contexts from which specific policies may be generated given context. Empirically, we expect each...

Generative Policy-based Models aim to enable a coalition of systems, be they devices or services adapt according contextual changes such as environmental factors, user preferences and different tasks whilst adhering various constraints regulations directed by managing party the collective vision coalition. Recent developments have proposed new architectures realize potential GPMs but complexity systems their associated requirements increases, there is an emerging requirement scenarios...

Logic-based machine learning aims to learn general, interpretable knowledge in a data-efficient manner. However, labelled data must be specified structured logical form. To address this limitation, we propose neural-symbolic framework, called Feed-Forward Neural-Symbolic Learner (FFNSL), that integrates logic-based system capable of from noisy examples, with neural networks, order unstructured data. We demonstrate the generality FFNSL on four classification problems, where different...