Over the past years deep learning has brought a revolution in the area of artificial intelligence (AI), producing remarkable results in a variety of application domains including computer vision, natural language processing, speech recgonition, robotics, and clinical decision making. Despite the success of deep learning over a wide spectrum of real-world tasks, there is no doubt that many of the problems that are really at the core of AI are far from being solved. Reasoning, namely taking pieces of information, combining them together, and using these to draw logical conclusions or devise new information, is not a general-purpose capability for modern AI. Imagine an airplane passenger sitting in an exit row, studying the emergency guide, which is often a combination of images and text. Their brain combines visual and textual information in order to infer the intended message -- open the door in the unlikely event of an emergency. A computer system seeing the same document would first employ an image recognition model to scan the image. An Optical Character Recognition (OCR) system would read the text, and a third system would correlate the image and text to understand the complete picture. Although the fundamental principles of analyzing the world around us and the approach a machine takes to process complex information are both based on breaking down the data to its core elements, humans are instinctively better at correlating and integrating information from different modalities, and re-using previously acquired experience and expertise to transfer it to radically different challenges and domains. Today's neural networks fail disastrously when exposed to data outside the distribution they were trained on, overly adhere to superficial and potentially misleading statistical associations instead of learning true causal relations, are unable to reason on an abstract level, which makes it difficult to implement high-level cognitive functions, and are essentially black boxes with with respect to human understanding of their predictions. This fellowship aims to alleviate these deficiencies by developing a new class of neural network models which will demonstrate reasoning capabilities, a skill required to enhance many AI applications. Rather than relying on a monolithic network structure, we propose to assemble a network from a collection of more specialized modules, making use of an explicit, modular reasoning process, which allows for differentiable training (with backpropagation) but without expert supervision of reasoning steps. We will develop a theoretical framework which characterizes what it means for neural network models to reason, design various reasoning modules, and showcase their practical importance in applications which understand requests and act on them, process and aggregate large amounts of data (e.g., from multiple modalities), make generalizations (e.g., robots cannot be pretrained on all possible scenarios they might encounter), deal with changing situations and causality, manifest creativity (e.g., in writing a story or a poem), co-ordinate various agents (e.g., in game playing), and are able explain their predictions and decisions. The proposed Fellowship will have a transformative effect on AI theory and practice. It sets an ambitious agenda which unifies multiple strands of AI research, bridging the gap between the neural and symbolic views of AI and integrating their complementary strengths. It will provide the means for developing a UK skill base in AI, and wil have wide ranging impact in academia, industry, the UK economy, and society e.g., , by embedding AI in many domains of daily life and rendering tools such as neural networks more explainable.

Recent reports from the Royal Society, the government Cybersecurity strategy, as well as the National Cyber Security Center highlight the importance of cybersecurity, in ensuring a safe information society. They highlight the challenges faced by the UK in this domain, and in particular the challenges this field poses: from a need for multi-disciplinary expertise and work to address complex challenges, that span from high-level policy to detailed engineering; to the need for an integrated approach between government initiatives, private industry initiatives and wider civil society to tackle both cybercrime and nation state interference into national infrastructures, from power grids to election systems. They conclude that expertise is lacking, particularly when it comes to multi-disciplinary experts with good understanding of effective work both in government and industry. The EPSRC Doctoral Training Center in Cybersecurity addresses this challenge, and aims to train multidisciplinary experts in engineering secure IT systems, tacking and interdicting cybercrime and formulating effective public policy interventions in this domain. The training provided provides expertise in all those areas through a combination of taught modules, and training in conducting original world-class research in those fields. Graduates will be domain experts in more than one of the subfields of cybersecurity, namely Human, Organizational and Regulatory aspects; Attacks, Defences and Cybercrime; Systems security and Cryptography; Program, Software and Platform Security and Infrastructure Security. They will receive training in using techniques from computing, social sciences, crime science and public policy to find appropriate solutions to problems within those domains. Further, they will be trained in responsible research and innovation to ensure both research, but also technology transfer and policy interventions are protective of people's rights, are compatible with democratic institutions, and improve the welfare of the public. Through a program of industrial internships all doctoral students will familiarize themselves with the technologies, polices and also challenges faced by real-world organizations, large and small, trying to tackle cybersecurity challenges. Therefore they will be equipped to assume leadership positions to solve those problems upon graduation.

The UKRI CDT in Artificial Intelligence (AI) for Healthcare will be the world's leading centre for PhD training of the next-generation innovators in AI applied to Healthcare. There is a unique role for AI in healthcare by providing more accurate decisions faster while reducing cost and suffering across society. AI in healthcare needs and drives current AI research avenues such as interpretable AI, privacy-preserving learning, trust in AI, data-efficient learning and safety in autonomy. These are key due to the immediate impact on life and health for users depending on AI for healthcare support. Healthcare applications require many AI specialists that can apply their skills in this heavily regulated domain. To address this need, we propose to train in total 90+ PhD students including 16 clinical PhD Fellows in five cohorts of 18+ PhDs, which will establish a new generation of cognitively diverse AI researchers with backgrounds ranging from computer science, psychology to design engineering and clinical medicine. The CDT focus areas arise from our early engagement in AI research and collaboration with clinicians, partnered technology companies and patient organisations, reflecting the healthcare areas of the UK industrial strategy. The Centre is grouped into 4 complementary healthcare themes and 4 cross-cutting AI expertise streams. The 4 healthcare themes are: (1) Productivity in Care: making healthcare provision more efficient and effective by increasing the productivity of doctors and nurses; (2) Diagnostics & Monitoring: developing AI-based diagnostics & monitoring that can detect disease earlier and monitor health with more precision; (3) Decision support systems: AI-based decision support systems that will support e.g. freeing up doctors' time to focus on the patient or can accelerate the development of novels drugs and treatments and empowering patients to be active agents within the decision-making by explaining, and (4) Biomedical discovery: driven by AI that accelerates drug discovery and linking genome, microbiome and environment data to discover novel disease mechanisms and treatment pathways. The themes are linked by 4 cross-cutting AI expertise streams: a. Perceptual AI technology enables to perceive, structure, and recognise from sensory data clinically relevant information. b. Cognitive AI technology mimics the reasoning, i.e. cognitive process, of healthcare specialists. c. Assistive AI technology supports clinicians with decision making as well as patients directly d. Underpinning AI technologies are driving factors for clinical and patient-focused AI innovations and will be enabling AI methodologies to operate beyond the currently possible. Our unique cohorts will benefit from an integrated training program and co-creation process with industry and patient organisations. PhD training is split into three phases that provide underpinning skill training (Foundation phase), research training (Research Phase) and finally drive PhD impact (Impact phase). During the Impact phase, the students will either (1) commercialising their research through a mentored start-up route (incubator partners), (2) deploying their technology in a clinical trial (two NIHR biomedical research centre (BRC) partners), or (3) testing their work in person through an NHS honorary contract (three NHS trusts as partners). Bespoke training will be created, such as AI bias & ethics, security, trust, inclusivity, differential privacy, transparency, accessibility and usability, service design, global inclusivity, healthcare treatments, clinical statistics and data regulation, Healthcare technology regulation, and technology commercialisation. We offer an exit Strategy (month 9-12) through a master's degree. The centre will place special emphasis on research that explores diversity in AI for healthcare research, including services to underserved communities and minority-specific care requirements.

Research Area: ART-AI is a multidisciplinary CDT, bringing together computer science, social science and engineering so that its graduates will be specialists in one subject, but have substantial training and experience in the others. The ART-AI management team brings together research in AI, HCI,politics/economics, and engineering, while the CDT as a whole has a team of >40 supervisors across seven departments in three faculties and the institutes for policy research (IPR) and for mathematical innovation (IMI). This is not a marriage of convenience: many CDT members have experience of interdisciplinary working and together with CDT cohorts and partners, we will create accessible, transparent and intelligible AI, driven by ethical and responsible principles, to address issues in, for example, policy design and political decision-making, development of trust in AI for humans and organisations, autonomous systems, sensing and data analysis, explanation of machine decision-making, public service design, social simulation and the ethics of socio-technical systems. Need: Hardly a day passes without a news article on the wonders and dangers of AI. But decisions - by individuals, organisations, society and government - on how to use or not use AI should be informed and ethical. We need policy experts to recognise both opportunities and threats, engineers to extend our technical capabilities, and scientists to establish what is tractable and to predict likely outcomes of policies and innovations. We need mutually informed decisions taking account of diverse needs and perspectives. This need is expressed in measured terms by a slew of major reports (see Case for Support) and Commons and Lords committees, all reflecting the UKCES Sector Insights (Evidence report #92, 2015) prediction of a need by 2022 for >0.5M additional workers in the digital sector against just a third of that number graduating annually. To realise the government vision for AI (White Paper), a critical fraction of those 0.5M workers need to be leaders and innovators with in-depth scientific and technical knowledge to make the right calls on what is possible, what is desirable, and how it can be most safely deployed. Beyond the UK, a 2018 PwC report indicates AI will impact ~10% of jobs, or ~326 million globally by 2030, with ~33% in high-skill jobs across most economic sectors. The clear conclusion is a need for a significant cadre of high-skill workers and leaders with a detailed knowledge of AI, an understanding of how to utilise it, and its political, social and economic implications. The ART-AI is designed to deliver these in collaboration and co-creation with stakeholders in these areas. Approach: ART-AI will produce interdisciplinary graduates and interdisciplinary research by (i) exposing its students to all three disciplines in the taught elements, (ii) fostering development of multi-discipline perspectives throughout the doctoral research process, and (iii) establishing international and stakeholder perspectives whilst contributing to immediate, real-world problems through a programme of visiting lecturers, research visits to leading institutions and internships. The CDT will use some conventional teaching, but the innovations in doctoral training are: (i) multi-disciplinary team projects; (ii) structured and facilitated horizontal (intra-cohort) peer learning and vertical (inter-cohort) mentoring, and in the interdisciplinary cross-cohort activities in years 2-4; (iii) demonstrated contextualisation of the primary discipline research in the other disciplines both at transfer (confirmation) at the end of year 2 and in the final dissertation. Each student will have a primary supervisor from their main discipline, a co-supervisor from at least one of the other two, and where appropriate, one from a CDT partner, reflecting the interdisciplinarity and co-creation that underpin the CDT.

Our vision for this Transition Award is to leverage and combine key emerging technologies in Artificial Intelligence (AI) and Engineering Biology (EB) to enable and pioneer a new era of world-leading advances that will directly contribute to the objectives of the National Engineering Biology Programme. Realisation of the benefits of Engineering Biology technologies is predicated on our ability to increase our capability for predictive design and optimisation of engineered biosystems across different biological scales. Such a scaled approach to Engineering Biology would serve to significantly accelerate translation of scientific research and innovation into applications of wide commercial and societal impact. Synthetic Biology has developed rapidly over the past decade. We now have the core tools and capabilities required to modify and engineer living systems. However, our ability to predictably design new biological systems is still limited, due to the complexity, noise, and context dependence inherent to biology. To achieve the full capability of Engineering Biology, we require a change in capacity and scope. This requires lab automation to deliver high-throughput workflows. With this comes the challenge of managing and utilising the data-rich environment of biology that has emerged from recent advances in data collection capabilities, which include high-throughput genomics, transcriptomics, and metabolomics. However, such approaches produce datasets that are too large for direct human interpretation. There is thus a need to develop deep statistical learning and inference methods to uncover patterns and correlations within these data. On the other hand, steady improvements in computing power, combined with recent advances in data and computer sciences have fuelled a new era of Artificial Intelligence (AI)-driven methods and discoveries that are progressively permeating almost all sectors and industries. However, the type of data we can gather from biological systems does not match the requirements for off-the-shelf ML/AI methods and tools that are currently available. This calls for the development of new bespoke AI/ML methods adapted to the specific features of biological measurement data. AI approaches have the potential to both learn from complex data and, when coupled to appropriate systems design and engineering methods, to provide the predictive power required for reliable engineering of biological systems with desired functions. As the field develops, there is thus an opportunity to strategically focus on data-centric approaches and AI-enabled methods that are appropriate to the challenges and themes of the National Engineering Biology Programme. Closing the Design-Build-Test-Learn loop using AI to direct the "learn" and "design" phases will provide a radical intervention that fundamentally changes the way that we design, optimise and build biological systems. Through this AI-4-EB Transition Award we will build a network of inter-connected and inter-disciplinary researchers to both develop and apply next-generation AI technologies to biological problems. This will be achieved through a combination of leading-light inter-disciplinary pilot projects for application-driven research, meetings to build the scientific community, and sandpits supported by seed funding to generate novel ideas and new collaborations around AI approaches for real-world use. We will also develop an RRI strategy to address the complex issues arising at the confluence of these two critical and transformative technologies. Overall, AI-4-EB will provide the necessary step-change for the analysis of large and heterogeneous biological data sets, and for AI-based design and optimisation of biological systems with sufficient predictive power to accelerate Engineering Biology.