Artificial intelligence (AI) brings without any doubt huge opportunities to optimize efficiency of every industrial application and is a key point of Industry 4.0. The deployment of various sensors in factories, also called Industrial Internet of Things (IIoT) can either help workers in charge of machine maintenance by detecting abnormal behaviours, thus preventing machine breakdown, or help to localize objects or persons in such complex environments. AI algorithms probably represent the best solution to cope with the huge amount of data provided by sensors, but their complexity is also a severe drawback and the processing is mainly centralized. Energy is crucial for IIoT, since the more sensors are deployed, the more difficult it becomes to ensure sufficient energy, as batteries would need to be recharged more frequently. Moving the processing closest as possible to the sensors would avoid energy hungry transmissions of data. Most of the latter is indeed useless, since AI algorithms need to be fed with descriptors more than raw data. To further enhance energy efficiency of Edge AI, LIGHT-SWIFT aims at proposing a new methodology to reduce the complexity of AI algorithms, paving the way for sustainable smart sensors in Industry 4.0. This methodology will be applied to sound sensor nodes able to detect unusual situations, either in machine behaviour but also in the general context of the factory. In case of emergency, the system may have to cope with massive amounts of additional data, entailing a crucial need for extremely reliable high data rate transmissions, despite the limited spectrum resources. The methodology of LIGHT-SWIFT project will therefore be applied to the wireless transmissions themselves, to optimize the radio resource access, while achieving the best possible energy efficiency. To reach this goal, our project will leverage a well-balanced consortium composed of two academic partners, IRISA and NII, that work respectively on energy efficient wireless sensor networks and edge AI for wireless communications, the Small and Medium-sized Enterprise (SME) Wavely specialized in sound event detection for IIoT, and one of the largest telecommunications operating companies in the world, NTT, with applications in Industry 4.0.

Due to recent advances in speech and language processing capabilities, humans can today interact with intelligent technology by using their voice. The number of voice-enabled devices is growing exponentially and consumers are adapting quickly to using their voice as a natural means of interaction. Today’s voice interfaces can project synthetic speech signals that are of such quality that they are (near-to) indistinguishable from that of human speech. They are also capable of responding to requests or commands issued in the form of entirely natural, conversational speech and can even recognise or identify the user from their speech alone. This project concerns the components of voice interfaces that relate to, or impact upon the notion of speaker/voice identity. They include speech generation and speaker recognition technologies. Speech generation technologies are components of a voice interface that aim to produce a natural human voice. They include speech synthesis and voice conversion technologies, both of which have the capacity to produce speech signals that are representative of a specific speaker identity. Speaker recognition technologies are the components of a voice interface that aim to determine or verify the identity of a human speaker. In some senses, speech generation and speaker recognition technologies have potentially conflicting objectives. Speech generation technologies aim to produce human speech artificially whereas speaker recognition technologies aim to verify the authenticity of human speech and a claimed identity. Speaker recognition systems may thus be used to help train speech generation system. As a consequence, artificially generated speech then has the potential to fool recognition systems. Herein lies the conflicting objective. A second conflict stems from the use of speaker recognition technologies when a speaker may not wish to be identified or tracked. In order to protect the right to anonymity, de-identification solutions are then required in order to supress identity information from a speech while retaining linguistic information (the message). The study of speaker/voice identity in all three aspects of speech generation, speaker recognition and privacy are closely intertwined. The VoicePersonae project will bridge the technical gap between the fields linked to voice identities and will (a) advance speaker identity modelling, (b) improve the security and robustness of biometric speaker recognition and (c) invent new solutions to conserve speaker privacy. For the accurate modelling of voice identities, required for new applications such as personal avatars and in the health domain, VoicePersonae will unify several classical multi-speaker speech generation tasks, that is, multi-speaker speech synthesis, voice conversion and speech enhancement. VoicePersonae will harness speaker recognition technologies in order to achieve this goal. In order to improve the robustness of speaker recognition to the security threats presented by advances in speech generation, VoicePersonae will also deliver advances in anti-spoofing. This work will be undertaken assuming that fraudsters are aware of anti-spoofing technologies and hence attempt to spoof not only biometric recognition systems but also anti-spoofing systems. Finally, VoicePersonae will deliver speaker anonymization capabilities in order to provide for speaker privacy. In order to fuel progress in this area, VoicePersonae will organise the first speech anonymization and re-identification challenge.

Our aim here is to enhance reliability of AI in society by implementing realtime compliance mechanism for legal and ethical norms. Our contribution is to build a compliance mechanism by considering legal norms as hard constraints which must be satisfied and ethical norms as soft constraints which should be satisfied as much as possible. This combination of multiple norm compliance has not been investigated and is one of the novel parts of the project We retain realtime scalability by introducing a partial evaluation mechanism along the execution sequence of an AI agent that checks the legal norms and a speculative computation that checks ethical norms with multiple possible sequences of comparison of these soft norm. There are many researches working on offline compliance check of norms whereas there are few researches working on online compliance check. In addition, to our knowledge, these online compliance mechanisms only check the violation of the hard norms while they don’t consider soft norms violation since it needs online norm revision. We investigated here such online belief revision method in soft constraints called ‘’speculative computation’’ and we will apply this method to soft norm revision. As far as we know, this is the first attempt to formalize online norm revision. Japanese team has been long working on legal reasoning and offline compliance mechanism of legal norms and proposed a legal representation language called PROLEG (PROLOG based LEGal reasoning system). French team has been working on formalizing ethics in logic and given a rigorous framework using Event Calculus which represents temporal behavior of AI agents to represent various variations of ethical norms. German team has been working on knowledge representation issues such as aspect- oriented (metadata) knowledge modelling and reasoning with such aspect metadata scopes (scoped reasoning). They have developed tools and standards to represent rules (norms). We expect to develop a unified system of handling various norms such as legal norms and ethical norms simultaneously. Their tools and standards will be used to describe various norms. Therefore, the combination of three teams is essential to achieve our goal. Moreover, we also expect that combining legal norms and ethical norms will provoke interesting interactions between each other and it leads to new research topics to understand normative reasoning more deeply. If we succeed in this project, we expect that AI will be more reliable entity and a good partner with humans in the upcoming years.

Do you use mobile or web apps or have Internet of Things devices on your person, in your home or workplace? Have you thought about who developed the software that drives these apps and devices, what was their understanding of cyber security, how did they make design decisions that impact the cyber security of the resulting software, and what factors influenced their behaviour and design choices? Or perhaps you are one of the masses exploiting app development platforms and easy-to-program hardware devices such as Arduino and Raspberry Pi to develop applications and deploy them for personal use or distribute them to millions of people around the world? How do you make cyber security decisions when you write software? Do you consciously think about the security implications of your design choices, or are there other factors that are more critical? What will help you achieve your goals from the software that you are developing while ensuring that it is not vulnerable to attacks by malicious actors? This project aims to develop a deep foundational understanding of these issues. We recognise that developing software is no longer the preserve for the select few with deep technical skills, training, and knowledge. A wide range of people from diverse backgrounds are increasingly developing software for mobile and web apps and for programmable consumer devices. This diversity of developers is at the heart of many innovations in the digital economy. The software they produce can be, and is, deployed across systems embedded in many aspects of human activity, and is used by a global user base. However, little is currently understood about the security behaviours and decision-making processes of 'the masses' engaged in software development. We refer to these masses by the pseudonym 'Johnny' - based on a seminal work by Whitten and Tygar where they highlighted the challenges faced by Johnny, the prototypical user of encryption. In this project we aim to tackle the challenges faced by Johnny in a contemporary setting beyond encryption. We focus on the Johnnys with diverse backgrounds, know-how and cyber security expertise who can, and are, developing software used, potentially, by millions worldwide. Drawing on a research team of experts in cyber security, software engineering, and psychology, our aim in this project is to conduct empirically-grounded research to better understand the security implications of Johnny's behaviours and practices and develop effective support for secure software development by Johnny. We propose to achieve this by uncovering and characterising the security vulnerabilities that Johnny tends to introduce, by analysing how and why these vulnerabilities are introduced, and by identifying and evaluating a range of interventions to improve Johnny's security behaviours during software development. We will do this in collaboration with eminent international research partners, drawn from leading research and practitioner organisations around the world. This project will be the first to study the inter-relationship between the cognitive and social processes that shape Johnny's cyber security decisions, their impact on the security of the resultant software and the novel interventions that may steer Johnny towards more effective cyber security decisions during software development.

Do you use mobile or web apps or have Internet of Things devices on your person, in your home or workplace? Have you thought about who developed the software that drives these apps and devices, what was their understanding of cyber security, how did they make design decisions that impact the cyber security of the resulting software, and what factors influenced their behaviour and design choices? Or perhaps you are one of the masses exploiting app development platforms and easy-to-program hardware devices such as Arduino and Raspberry Pi to develop applications and deploy them for personal use or distribute them to millions of people around the world? How do you make cyber security decisions when you write software? Do you consciously think about the security implications of your design choices, or are there other factors that are more critical? What will help you achieve your goals from the software that you are developing while ensuring that it is not vulnerable to attacks by malicious actors? This project aims to develop a deep foundational understanding of these issues. We recognise that developing software is no longer the preserve for the select few with deep technical skills, training, and knowledge. A wide range of people from diverse backgrounds are increasingly developing software for mobile and web apps and for programmable consumer devices. This diversity of developers is at the heart of many innovations in the digital economy. The software they produce can be, and is, deployed across systems embedded in many aspects of human activity, and is used by a global user base. However, little is currently understood about the security behaviours and decision-making processes of 'the masses' engaged in software development. We refer to these masses by the pseudonym 'Johnny' - based on a seminal work by Whitten and Tygar where they highlighted the challenges faced by Johnny, the prototypical user of encryption. In this project we aim to tackle the challenges faced by Johnny in a contemporary setting beyond encryption. We focus on the Johnnys with diverse backgrounds, know-how and cyber security expertise who can, and are, developing software used, potentially, by millions worldwide. Drawing on a research team of experts in cyber security, software engineering, and psychology, our aim in this project is to conduct empirically-grounded research to better understand the security implications of Johnny's behaviours and practices and develop effective support for secure software development by Johnny. We propose to achieve this by uncovering and characterising the security vulnerabilities that Johnny tends to introduce, by analysing how and why these vulnerabilities are introduced, and by identifying and evaluating a range of interventions to improve Johnny's security behaviours during software development. We will do this in collaboration with eminent international research partners, drawn from leading research and practitioner organisations around the world. This project will be the first to study the inter-relationship between the cognitive and social processes that shape Johnny's cyber security decisions, their impact on the security of the resultant software and the novel interventions that may steer Johnny towards more effective cyber security decisions during software development.