Artificial intelligence (AI) is transforming our societies, but the more it proliferates, the higher the customer demands for functionality and efficiency (most notably energy). Thus, as time progresses the limitations of statistical learning-based AI that has underpinned most AI work so far are beginning to naturally become more exposed. Tasks such as variable binding and manipulation, inductive reasoning and 1-shot learning, at which statistical learning is not as strong, suggest solutions in the sphere of abstract symbol processing AI. The commonly referenced 'next wave of AI' that is capable of such exploits (towards "strong AI") is likely to make extensive use of symbol processing capabilities and simultaneously demand a bespoke set of hardware solutions. The proposed project primarily addresses the issue of developing general-purpose (platform-level) hardware for precisely symbolic AI. The proposed project seeks to develop a memory module that features: a) an internal structure and b) in-memory computing capabilities that render it particularly suitable for symbolic processing-based artificial intelligence (AI) systems. Ultimately the project seeks to deliver: 1) Two microchip iterations prototyping the memory system. 2) A software environment (infrastructure) for easy programming and operation of the resulting microchips (includes simulation capabilities for proof-of-concept tests). 3) A demonstration of the memory cell operating together with a symbolic processor as an aggregate system. 4) A functioning set of starter applications illustrating the capabilities of the design. The overall effort is driven by a philosophy of co-optimising the memory across the entire trio of fundamental device components, symbolic AI mechanics and hardware design facets. Specifically: functionality in the proposed memory system will be pursued by: a) Designing a resistive RAM-based (ReRAM) memory unit where operation of the ReRAM devices and ReRAM tech specifications themselves are subservient to the specific operational goals of the memory system. b) Adapting the mathematical machinery of the system in order to map functional operations to hardware-friendly machine-code level operations: the stress is on hardware-friendliness, not mathematical elegance. This will be inextricably linked to the design of the memory's instruction set. c) Designing an architecture that runs the symbolic memory efficiently by using memory allocation techniques that maximise locality and making extensive use of power-gating. Simultaneously, implementation of a solid software stack infrastructure will enable efficient and fast prototyping and hypothesis testing. The cornerstone of the targeted project impact is to lay the foundations for launching an industrial-scale design effort towards hardware for symbolic AI. Hence the bulk of the effort is in chip design (prototype-based de-risking of the idea) and toolchain development (impact acceleration by lowering barriers to user uptake). Simultaneously, it is expected that the project will play a significant role in enhancing interest in symbol-level AI and very crucially, inducing interest in connecting symbolic AI with statistical learning one; thereby significant impact on knowledge is achieved. Finally, the increased in the capabilities of AI, as well as the transparency of decision-making (typically readily expressible via formal expressions or even in pseudo-natural language) offered by the symbolic approach promise to make a significant impact in enhancing acceptance of AI by society, providing a solid scientific foundation for certification processes (AI trust - broadening the scope of applications that accept an AI solution). With hardware available for this task, significant impact on productivity and quality of life is to be achieved. The project is self-contained and is designed to launch a much broader, sustainable effort, headed by the PI in this field.

Nerve agents are amongst the most deadly chemicals known to man and continue to pose a significant societal threat. Nerve agents function by inhibiting chemicals in the brain that affect the nervous system. Thus, nerve agents cause seizures in the brain which can lead to severe brain damage and even death. Several drugs are used as initial treatments for nerve agent poisoning, but these are sometimes ineffective and can themselves be harmful. In addition, the greater the delay between exposure to the nerve agent and the provision of treatment, for example on the battlefield, the lower the likelihood of the effectiveness of these treatments. Consequently, new and better treatment options are needed to protect against the effects of nerve agents. Current methods used for testing new drug effectiveness for the treatment of nerve agent poisoning are largely reliant on the use of rodents. Such experiments are slow and costly, and usually involve severe procedures, such as the surgical implantation of electrodes in the brain and exposure to nerve agents. There is a requirement to develop higher throughput methods for identifying novel treatments, that are also more ethically favourable than those currently available. The non-protected 4-days post-fertilisation (dpf) larval zebrafish could prove invaluable as they have been shown to be responsive to a range of seizure-inducing drugs and can be tested quickly and easily in large numbers. Scientists from the universities of Portsmouth and Exeter will build on previous NC3Rs-funded work to transfer a non-protected larval zebrafish seizure model to Dstl, where the methods can be used to identify novel treatments for nerve agent poisoning. Dr Parker is a zebrafish behavioural expert, and will develop behavioural measures of seizures in 4dpf larvae. Many of the protocols were developed during Dr Parker's work on an NC3Rs project grant at Queen Mary, London (PI Caroline Brennan). Dr Winter is an expert in examining the brain during seizures in zebrafish using advanced imaging techniques, some of which have been developed during an ongoing NC3Rs studentship. His team will focus on developing approaches assessing seizure activity in the 4dpf zebrafish brain to understand model relevance for predicting effects in mammals. The end users at Dstl will utilise this approach for the identification and development of novel treatments for nerve agent poisoning. Dstl colleagues will promote the wider uptake of this approach, and the zebrafish as a model for assessing chemical toxicity, within the international defence research community. This approach could replace a significant number of rodents used in testing novel treatments against nerve agent toxicity, thus reducing overall rodent use by an estimated 75%. Limited rodent experiments would remain only for confirmatory purposes. Our approach could therefore prevent the yearly global use of at least 1500 rodents in these severe protocols. In addition to the direct replacement of rodents, the data generated in non- protected zebrafish larvae can also be used to refine remaining rodent studies to ensure that appropriate non-toxic doses are used. Refinement will also result from the identification and ruling out of any putative treatments with undesirable properties prior to escalation to rodent models. Dstl actively participates in a number of international research collaborations including bilateral arrangements with European countries and an important multinational agreement between the Australia, Canada, the UK and the USA, the CBR Memorandum of Understanding (MOU). Dr Kearn will use these arrangements and his position as UK lead for a predictive toxicology task under the CBR MOU to share data and methodologies from this project, champion its outputs and influence other Nations' programmes to encourage uptake of this technology.

RESEARCH CONTEXT An estimated 25,000 foreign fighters have been recruited to Islamic State (IS) from over 100 counties around the world. While shocking in itself, this reflects only the militant end-product of the radicalisation process. Before militancy emerges, IS supporters use mainstream social media in English to socialise, radicalise and recruit foreigners. For example, the fourth IS 'Clanging of the Swords' film racked up millions of views on video-sharing platforms, and created vast excitement among those who followed ISIS online. This use of online communications in English by supporters of Islamic State (IS) presents two unprecedented research opportunities: first, to predict the spread of extremism in English-speaking populations; and second, to understand how online communications shape people's understanding of IS and its opponents. However, currently there is no method to analyse longitudinal social media data to investigate the processes underlying the radicalisation of social media users. The proposed research provides a timely opportunity to refine methods, software and analytics to maximise the impact that psychology and computer science can make to countering extremism. The central research questions are how and why people develop allegiance to extremist groups through communicating online. To do this, we will develop the functionalities and capabilities of an existing software suite that can harvest and analyse large volumes of publicly-available Twitter data ('Tweets'). AIMS AND OBJECTIVES The central aim of this project is to develop conceptually-grounded social media analytics software to investigate the socialisation and radicalisation of mainstream users. As a by-product of this process, and secondary aim, we will establish how supporters of radical Islam use online communications in English to affect how their target audience defines both themselves and their opponents in relation to IS. APPLICATIONS AND BENEFITS This project will deliver a software tool that can analyse longitudinal big data. This software will be open source and freely available. This means that it will have the potential to make a big impact in social science. It will be applicable to, and usable by, any researcher or practitioner interested in analysing social media data. PROJECT TEAM Our project brings together a multi-disciplinary team of academics and will be steered by end-users, mediated by Defence Science and Technology Laboratory, whose expertise will enable us to maximise the usefulness and applicability of the software tools. On the academic side, we have assembled a team with interdisciplinary strengths in computer science, psychology and social media data analysis to understand the needs of social scientists and successfully develop technology that will become fully embedded in social science and social media research.

Nuclear magnetic resonance (NMR - the technology behind magnetic resonance imaging or MRI as used in medical scanning) offers enormous opportunity for materials detection and characterisation in the "real-world": that is open-access, electromagnetically unshielded applications outside of the laboratory where it is necessary to assess the quality, state or presence of materials. These applications include: the minimisation of delays to allow concrete to cure during construction or the assessment of building degradation in the built environment; managed forestry in order to decide which trees to fell or to minimise energy consumption from timber drying during processing; or illicit material detection at secure locations such as airports; and down bore-hole logging for oil and gas well exploration. However, save oil and gas exploration, where the earth's crust provides a natural electromagnetic shield, external radio frequency interference (RFI) pick-up restricts materials detection limits using affordable and practical light-weight, low-field strength magnets and hence inhibits the technology being taken up widely by industry. The problem is unwanted pick-up of radio signals such as aeronautical communications and amateur radio. The pick-up dominates and masks the weak NMR signals coming from materials to be inspected in open-access detectors. It is the purpose of this proposal to address RFI pick-up head on. We propose a new method to eliminate pick-up called active RFI suppression. We will build a technology demonstrator for active RFI suppression in an open-access NMR system. Our target is a greater than 10 times improvement in the signal-to-noise ratio, sufficient to enable take-up in construction, forestry and homeland security. The project takes advantage of recent rapid developments in real time signal processing and radio frequency engineering emanating from the mobile communications industry that enable modern devices to process radio signals in real time at the carrier frequency. With these advances we are able to measure the RFI pick-up independently of the NMR signal. Hence we can eliminate the pick up from the NMR signal. The team assembled for the project uniquely comprise experts in NMR techniques and applications and applied radio frequency electronic engineering. The project builds on an RFI characterisation and software feasibility study carried out jointly with Dstl during 2014 and our own experience of using NMR in construction and forestry both in the laboratory and outside in the "real-world" on construction sites and in forests. Dstl remain involved and contribute through support for an engineering doctorate research student, the provision of technical hardware on which to build the demonstrator and detailed knowledge of security applications.

Airborne crop diseases pose a serious threat to food security and are responsible for devastating loss of yield and over-reliance on pesticides. Early detection enables farmers to take preventative action, drastically reducing damage and cost. Current detection regimes often rely on expert identification of the pathogen from plant damage. More recently, other molecular techniques have emerged. However, these methods suffer the same problems - being specific for a single species and a need for relatively large quantities of pathogenic material. Recently, TGAC has been working on an approach dubbed Air-seq that seeks to identify pathogens through sequencing of biological particles present in air. This overcomes both problems associated with current techniques as it is unbiased (not limited by species) and requires very small quantities of material. Our ultimate aim is to put sample collection, sequencing and analysis in a single box that can be deployed in the field. Key to success is a compact sequencing technology and this has recently emerged in the form of Oxford Nanopore Technologies' (ONT) MinION. The MinION is a new compact, low-cost sequencing technology that offers long reads (thousands of bases of DNA) and a streamed mode of operation enabling analysis of data as it is generated. These attributes make it ideally suited to in-field use. However, part of the process of generating sequencing data involves converting an electrical signal from the DNA sensing pore into a sequence of bases (letters) and this is performed via an internet 'basecalling' service. For in-field deployment, this is unsatisfactory, as we cannot rely on high speed, reliable data connections. We believe a completely new approach is required in which we utilise the raw signal data in order to identify species, instead of searching against basecalled sequence. In this project, we will develop a tool that searches Nanopore signal data looking for the characteristic signal traces of pathogens of interest, building up a report on abundance levels in the process.