'Musically Embodied Machine Learning' is an investigation into the musically expressive potential of machine learning (ML) when embodied within physical musical instruments. It proposes 'tuneable ML', a novel approach to exploring the musicality of ML models, when they can be adjusted, personalised and remade, using the instrument as the interface. Moving forward from the static preset models used in today's instruments, musicians playing instruments with a tuneable approach will be able to customise the ML models within their instruments, adapting to personal needs and varying situations, just as one might change the strings or pickups on an electric guitar, reconfigure modules in a synthesiser, or retune a set of drums for a particular performance. ML has been highly successful in the broader cultural and technical landscape, in allowing us to build novel creative tools for musicians. For example, generative models that bring new approaches to sound design, or models that allow musicians to build complex, nuanced mappings with musical gestures. These instruments offer new forms of creative expression, because they are configurable in intuitive ways using data that can be created by musicians. They can also offer new modes of control, with techniques such as latent space manipulation. Currently, to train a ML model, standard practice is to collect data (e.g sound or sensor data), create and pre-test the model within a data science environment, before testing it with the instrument. This distributed approach creates a disconnection between the instrument and the machine learning processes. With ML embodied within an instrument, musicians will be able to take a more creative and intuitive approach to making and tuning models, that will also be more inclusive to those without expertise in ML. This embodied approach to ML fits with wider views in the philosophy of artificial intelligence, on how we need to situate and embody models within the real world to improve them. Musicians can get the most value from ML if the whole process of machine learning is accessible; there are many creative possibilities in the training and tuning of models, so it's valuable if the musician can have access to the curation of data, curation of models, and to methods for ongoing retuning of models over their lifetime. We have reached the point where ML technology will run on lightweight embedded hardware at rates sufficient for audio and sensor processing. This opens up innumerable additions to our electronic, digital, and hybrid augmented acoustic instruments. Our instruments will contain lightweight embedded computers with ML models that shape key elements of the instruments behaviour, for example sound modification or gesture processing, responding to sensory input player and/or environment. This project will demonstrate how Tuneable ML creates novel musical possibilities, as it allows to create self-contained instruments, that can evolve independently from the complex data science tools conventionally used for ML. The project asks how instruments can be designed to make effective and musical use of embedded ML processes, and questions the implications for instrument designers and musicians when tunable processes are a fundamental driver of an instrument's musical feel and musical behaviour. With the speed of modern AI, it is clear that new instruments will emerge with tuneable ML, and is vital that we have a nuanced understanding of them, through experimental use in the hands of musicians, developers and designers. That way, we will support the design of future instruments, and offer new insights into our creative workflow with ML tools.

Model checkers are tools which can automatically verify that a system behaves as intended. Because this is done on a model of a system, design flaws are identified before the system is actually built, thus saving time, money, and even loss of lives (e.g. systems that control the navigation of airplanes). Here we focus on real-time model-checking, which deals with systems where timing constraints are important.A (real-time) model-checker accepts a model represented as a network of timed automata. Timed automata are a graph-like notation where nodes represent states in the system behaviour, and arcs between nodes represents actions. Both the time when actions can be performed, and the amount of time the system may remain in a given state, can be constrained. The second input to model-checkers is a logic formula, describing the correctness property we wish to check (e.g, that a certain undesired event will never occur). The logic is usually known as the requirements logic.Model-checking is one of the triumphs of theoretical computer science research, with a large number of successful applications in the commercial sector. This is because model-checkers can now automatically verify properties which, in the past, required experts to develop complex proofs by hand. Among real-time model-checkers, Uppaal is the most extensively applied. The success of Uppaal in bridging the gap between academic research and industrial application is impressive. However, the approach still has a number of significant limitations:1. Timelocks. These are degenerate states in which time is unable to pass and cannot, in the general case, be detected. Timelocks arise because of the way in which the passage of time is modelled in timed automata semantics. Of course, physical systems cannot stop time. However, the verification of properties by model-checkers is (for reasons that I will not address here) only meaningful for timelock-free models. If a model contains a timelock, then the user cannot have complete confidence in the verification results. For example, Uppaal may report that a bad event never occurs, unaware that the event may indeed occur, but cannot be detected because the model stops time before this happens. Dangerously, this undetected bad event may still be present when the system is built.2. Expressiveness of Requirements Logic. Uppaal is very efficient. But to achieve this, the designers had to limit the kind of properties which can be written (i.e they had to restrict the requirements logic). As a result, many properties that one would like to verify cannot be checked with Uppaal, or are difficult to express (and so it is easier to make mistakes when trying to capture the meaning of a given property).The proposed research will address these limitations and thereby significantly improve the applicability of real-time model-checking in general, and Uppaal, in particular. This will be done as follows.1. Building from existing related work on the subject I will develop techniques and tools to prevent or detect timelocks in timed automata specifications (these are based on the structure of the timed automata).2. I will integrate research on choppy logics (more expressive than Uppaal's logic) with test automata approaches (a different way to express properties, which Uppaal can handle efficiently) to enlarge the class of properties that Uppaal can model-check, without compromising its performance. 3. I will undertake a set of demanding case studies to evaluate our Uppaal extensions (e.g. control systems, communication protocols, sensor networks, etc.).4. Finally, I will feed the results of our research into fields of computing for which symbolic real-time model checking is critical.

The aim of this project is to create a collaborative international network to bring together researchers from two disciplines - economic geography/regional studies and organisational/workplace learning - who have been exploring issues related to learning, innovation and economic development from different theoretical perspectives. The aim is to generate new insights by exploiting synergies between the two, and the project will address two issues: (i) how the learning and skills of individuals is linked to collective learning by firms in different kinds of production-consumption networks, and (ii) how learning and innovation takes place across geographical, organisational and professional boundaries. These issues are recognised as being increasingly important to economic growth in the UK and elsewhere in Europe. Researchers in the area of economic development have noted that successful firms increasingly innovate through networks that stretch across regional and national boundaries, and that they combine knowledge and expertise from different professional and technical areas. In this context they have recognised that they need new conceptual tools to analyse the actual learning processes by which actors who are located in different places work together, and/or draw on different knowledge bases, to create new goods and services. Thus they have begun to engage with the work of researchers in the area of workplace and organisational learning who are developing new conceptualisations of the ways in which people learn to use new knowledge and tools between professional or technical communities to address the challenges they face in common, and recontextualise and reconfigure their practices in innovative ways. This project will consolidate and expand this nascent interdisciplinary collaboration. By generating new insights into these issues, the project will assist policymakers to develop/fine-tune policies in the areas of innovation, economic development and skills. The network will consult and engage with policymakers, practitioners and private sector actors throughout the project to disseminate emerging findings and obtain input into new research proposals that will be developed from our work.

The Feedback Musicianship Network (FMN) responds to the need to fill current gaps in knowledge around feedback instruments; we need a common language to describe their complex behaviour, and better understandings of: luthiery in hybrid instruments, virtuosity, composition and notation techniques. The FMN brings stakeholders in feedback musicianship together to establish a new research agenda addressing these gaps, and to build a community hub. This will stimulate and guide future developments in this field, supporting a new generation of instruments and musical practices. Feedback instruments offer a radically different way of engaging with musical practice compared to traditional instruments. They are defined by recirculation of signals through the instrument, which give the instrument 'a life of its own'; the player must guide the instrument rather than controlling it. They possess 'a stimulating uncontrollability' (Ulfarsson, 2019). The use of musical feedback began in the 1950s. Now, a new generation of instruments are using hybrid digital/electronic/acoustic technologies to refine the behaviour of the feedback, creating entirely new musical experiences, and providing fertile areas for creative new instrument designs and modes of musical practice. An example is the Feedback Cello, an acoustic cello augmented with string pickups and exciters; the string signals pass through external effects, and return to the cello through the exciters. This creates a feedback loop which the player navigates by damping and stimulating the strings, or by controlling the external effects. This is a radically different way of playing the cello, effectively turning it into a new instrument. In order to support the next generation of these instruments, we need to advance our understanding of how to shape the behaviour of complex feedback loops, and how to design and build instruments which are essentially hybrids, mixing complex signal processing with traditional acoustic luthiery, and electromechanical transducers that link these two domains. We also need to gain better understanding of the culture surrounding these instruments. This research demands interdisciplinary approaches involving music, engineering, mathematics, philosophy, design and computer science. The FMN will bring these groups together, along with practicing artists and industry representatives, for workshops and symposia at three themed network meetings: (1) Design, Making and Innovation, Aalborg University Copenhagen, (2) Musicianship and Notation, Berlin, (3) Approaches to Signal Processing, University of Sussex. The network will also run two longitudinal activities linking the three meetings: (1) composition of a piece for feedback ensemble, (2) progress reports from musicians learning and developing feedback instruments. These meetings will enable the community to establish a future research agenda, stimulate new activity in instrument design supported by knowledge exchange, and map out creative practices in feedback musicianship in order to guide future cultural engagement. The FMN has a strong interdisciplinary set of confirmed participants, and is guided by a highly qualified advisory board. It will engage further participants through live streaming and archiving of network events. The FMN will disseminate research though three peer reviewed journal articles, the key output being a research review and future research roadmap. Another key output of the network will be a new online hub for feedback musicians; we aim for this to become a focal point for the community to support future developments. The network will engage with the public at four concerts, also available online. Through concerts, knowledge exchange, and online sharing, the network will create impact by engaging the wider public in feedback musicianship, stimulating the design of new instruments and artistic practices, and by creating new dialogues between researchers and the public

The UK and global research and development communities have made tremendous strides in electronic device prototyping. Platforms that support conventional electronics have become well established, and the emerging potential of printed electronics and related additive technologies is clear. Together these support fast and versatile prototyping of the form and function of digital devices that underpin novel interactive data-driven experiences, including the Internet of Things (IoT), wearable technologies and more. However, challenges remain to realise their full potential. Interactive devices prototyped in labs and makerspaces implement novel capabilities and materials which require holistic manufacturing capability beyond simulation of conventional electronics. Even for conventional bench designs, to make the transition from prototype to product they need to be suitably robust, safe, long-lived, performant and cost-effective to deliver value as products - whether as a series of one-off mass customised devices, low-volume batches, or mass-produced artefacts. Unfortunately, the transition from prototype to production is not a natural one for end users; many ideas with potential don't progress beyond the first few designs. Democratising access to device production is the key next step in underpinning scalability and entrepreneurship in digital systems. We propose a Network+ of universities, research organisations and commercial enterprises who share the common goal of improving the transition from prototyping to production of digital devices. The Pro2 community will build upon the design and fabrication expertise of its researchers and practitioners to facilitate a deep synthesis of established principles, techniques and technologies and develop new concepts that span computer science, engineering and manufacturing. We will complement the on-going global investment into a variety of 'digital manufacturing' topics - including the UK's Made Smarter initiative - by tackling the challenge of progressively and cost-effectively transitioning from unconventional and single digital device prototypes, through tens of copies that can verify a design and validate utility, to batch production of hundreds to thousands of units. In prototyping, as additive manufacture and printed electronics converge further, in unconventional fields such as soft robotics and 4D printing, we need to identify how to integrate and optimise tools into workflows that support digital behaviour across materials, scales and functionalities. In production, smoothing the path from one-off microcontroller prototypes to scale-up is a significant challenge, and requires new processes and tools as well as reconfiguration of business models and services. Our vision for 'organic scaling' from prototype to production will allow faster exploration and exploitation of these digital device concepts and applications. This will accelerate the adoption of IoT, the growth of new consumer electronics markets, and more generally underpin the data-driven digital transformation of many industries. It will enable new research directions, create new business opportunities and drive economic growth.