Lasers are a key enabling technology in countless areas of modern society, touching on our lives in terms of ubiquitous connectivity, data storage, healthcare, security, environmental monitoring, etc. Examples include telecommunications, where they are used to generate the information carrying optical signals that are transmitted along thin glass optical fibres, manufacturing, where they are used for welding and cutting materials, and medicine, where they are used for sensing blood oxygen levels, and precisely resecting tissues. For almost all laser applications, it is necessary to use the laser source in combination with another technology that directs or "steers" the laser light in the desired direction. In some cases, this technology can be "passive", as is the case with the glass optical fibres used in telecommunications. In other cases, the steering technology must be "active" to change the direction of the laser beam in time, as is the case with the rapidly moving mirror systems used in some laser cutting and laser imaging systems. Conventional active laser steering technologies are often costly, bulky, and fragile. One or more of these disadvantages makes them sub-optimal for many important applications, including laser imaging systems for automotive applications, space-based laser communications systems, and drone-based remote sensing systems. To address this, there is currently a global drive to develop fully integrated solid-state beam-steering technologies, where the laser light is steered without the use of any physically moving components. Currently, however, even state-of-the-art solid-state laser beam steering systems have limited functionality, and do not meet the requirements of many real-world applications. In this project, we will exploit recent advances in two key integrated optical technologies - coherent Photonic Crystal Surface Emitting Laser (PCSEL) diode arrays and three-dimensional optical waveguide devices known as "integrated photonic lanterns" - to develop fully Integrated Solid-State Steerable Lasers (I-STEER) that can deliver agile beam steering in two dimensions and can, in principle, function at any diode laser wavelength. I-STEER will target the development of 900-mode PCSEL arrays, but will deliver the technological advances necessary to enable future PCSEL arrays (using commercial manufacturing facilities) that generate 10's of thousands of independently phase and ampltiude controllable coherent laser modes. A key aim of I-STEER is to enable denser PCSEL arrays, where the laser mode diameter is reduced to 20 microns (~20 wavelengths) and the centre-to-centre separation is reduced to ~50 microns (~50 wavelengths) - current PCSEL arrays exhibit 50 micron diameter laser modes with centre-to-centre separations of 400 microns. Unfortunately, even the ambitious spatial scales we are targeting mean that the PCSEL array will still be unsuitable for direct use as an optical phased array (OPA), since OPAs require very tightly packed wide angle emitters to achieve large angle/lobe free beam-steering. To address this, I-STEER introduces the fresh idea of using three-dimensional integrated optical waveguide transitions known as "integrated photonic lanterns" to adiabatically combine the PCSEL modes into a single highly multimode pattern of light, the spatial phase and amplitude properties of which can be directly controlled for beam steering via the PCSEL drive electronics. Through the I-STEER project, we aim to redefine the laser diode as an all-electronic integrated steerable light source enabling new functionally in countless applications including free-space optical communications and LiDAR. The generation of intellectual property and capability in this area will place the UK in a leading position with regards this strongly growing academic field, wealth generation through the creation of licensing and/or spin-outs, and in early adoption of UK based OEMs of this new technology.

Vision: In this fellowship, I aim to address a major challenge in the adoption of user-centred privacy-enhancing technologies: Can we leverage novel architectures to provide private, trusted, personalised, and dynamically- configurable models on consumer devices to cater for heterogenous environments and user requirements? Importantly, such properties must provide assurances for the data integrity and model authenticity/trustworthiness, while respecting the privacy of the individuals taking part in training and improving such models. Innovation and adoption in this space require collaborations between device manufacturers, platform providers, network operators, regulators, and the users. The objectives of this fellowship will take us far beyond the status-quo, one-size-fits-all solutions, providing a framework for personalised, trustworthy, and confidential edge computing, with ability to respect dynamic policies, in particular when dealing with sensitive models and data from the consumer Internet of Things (IoT) devices. In this fellowship, I aim to address these challenges by designing and evaluating an ecosystem where analytics from, and interaction with, consumer IoT devices can happen with trust in the model and authenticity, while enabling auditing and personalisation, hence pushing today's boundaries on all-or-nothing privacy and enabling new economic models. This approach requires designing for capabilities beyond the current trusted memory and processing limitations of the devices, and a cooperative dialogue and ecosystem involving service providers, ISPs, regulators, device manufacturers, and the end users. By designing our framework around the latest architectural and security features in edge devices, before they become commercially available, we provision for Model Privacy and a User-Centred IoT ecosystem, where service providers can have trust in the authenticity, attestability, and trustworthiness of the valuable models running on user devices, without the users having to reveal sensitive personal information to these cloud-based centralised systems. This approach will enable advanced and sensitive edge-based analytics to be performed, without jeopardising the individuals' privacy. Importantly, we aim to integrate mechanisms for data authenticity and attestation into our proposed framework, to enable trust in models and the data used by them. Such privacy-preserving technologies have the capacity to enable new form of sensitive analytics, without sharing raw data and thereby providing legal balancing capabilities that might enable certain sensitive (or currently unlawful) data analysis.

Highly available information networks are an increasingly essential component of the modern society. Targeted attacks are a key threat to the availability of these networks. These attacks exploit weak components in network infrastructure and attack them, triggering side-effects that harm the ultimate victim. Targeted attacks are carried out using highly distributed attacker networks called botnets comprising between thousands and hundreds of thousands of compromised computers. A key feature is that botnets are programmable allowing the attacker to adapt to evolve and adapt to defences developed by infrastructure providers. However current network infrastructure is largely static and hence cannot adapt to a fast evolving attacker. To design effective responses, a programmable network infrastructure enabling large-scale cooperation is necessary. Our research will create a new form of secure network infrastructure which detects targeted attacks on itself. It then automatically restructures the infrastructure to maximise attack resilience. Finally, it self-verifies whether global properties of safety and correctness can be assured even though each part of the infrastructure only has a local view of the world. Our research will examine techniques to collect and merge inferences across distributed vantage points within a network whilst minimising risks to user privacy from data-aggregation using novel privacy techniques. We make a start on addressing the risks introduced by programmability itself, by developing smart assurance techniques that can verify evidence of good intention before the infrastructure is reprogrammed. We set three fundamental design objectives for our design: (1) Automated and seamless restructuring of network infrastructure to withstand attacks aimed at strategic targets on the infrastructure. (2) A measurement system that allows dynamic allocation of resources and fine control over the manner, location, frequency, and intensity of data collected at each monitoring location on the infrastructure. (3) Assurance of safety and compliance to sound principles of structural resilience when infrastructure is reprogrammed. Our aim is to develop future network defences based on a smart and evolving network infrastructure.

Highly available information networks are an increasingly essential component of the modern society. Targeted attacks are a key threat to the availability of these networks. These attacks exploit weak components in network infrastructure and attack them, triggering side-effects that harm the ultimate victim. Targeted attacks are carried out using highly distributed attacker networks called botnets comprising between thousands and hundreds of thousands of compromised computers. A key feature is that botnets are programmable allowing the attacker to adapt to evolve and adapt to defences developed by infrastructure providers. However current network infrastructure is largely static and hence cannot adapt to a fast evolving attacker. To design effective responses, a programmable network infrastructure enabling large-scale cooperation is necessary. Our research will create a new form of secure network infrastructure which detects targeted attacks on itself. It then automatically restructures the infrastructure to maximise attack resilience. Finally, it self-verifies whether global properties of safety and correctness can be assured even though each part of the infrastructure only has a local view of the world. Our research will examine techniques to collect and merge inferences across distributed vantage points within a network whilst minimising risks to user privacy from data-aggregation using novel privacy techniques. We make a start on addressing the risks introduced by programmability itself, by developing smart assurance techniques that can verify evidence of good intention before the infrastructure is reprogrammed. We set three fundamental design objectives for our design: (1) Automated and seamless restructuring of network infrastructure to withstand attacks aimed at strategic targets on the infrastructure. (2) A measurement system that allows dynamic allocation of resources and fine control over the manner, location, frequency, and intensity of data collected at each monitoring location on the infrastructure. (3) Assurance of safety and compliance to sound principles of structural resilience when infrastructure is reprogrammed. Our aim is to develop future network defences based on a smart and evolving network infrastructure.

With the advent of deep learning and the availability of big data, it is now possible to train machine learning algorithms for a multitude of visual tasks, such as tagging personal image collections in the cloud, recognizing faces, and 3D shape scanning with phones. However, each of these tasks currently requires training a neural network on a very large image dataset specifically collected and labelled for that task. The resulting networks are good experts for the target task, but they only understand the 'closed world' experienced during training and can 'say' nothing useful about other content, nor can they be applied to other tasks without retraining, nor do they have an ability to explain their decisions or to recognise their limitations. Furthermore, current visual algorithms are usually 'single modal', they 'close their ears' to the other modalities (audio, text) that may be readily available. The core objective of the Programme is to develop the next generation of audio-visual algorithms that does not have these limitations. We will carry out fundamental research to develop a Visual Transformer capable of visual analysis with the flexibility and interpretability of a human visual system, and aided by the other 'senses' - audio and text. It will be able to continually learn from raw data streams without requiring the traditional 'strong supervision' of a new dataset for each new task, and deliver and distill semantic and geometric information over a multitude of data types (for example, videos with audio, very large scale image and video datasets, and medical images with text records). The Visual Transformer will be a key component of next generation AI, able to address multiple downstream audio-visual tasks, significantly superseding the current limitations of computer vision systems, and enabling new and far reaching applications. A second objective addresses transfer and translation. We seek impact in a variety of other academic disciplines and industry which today greatly under-utilise the power of the latest computer vision ideas. We will target these disciplines to enable them to leapfrog the divide between what they use (or do not use) today which is dominated by manual review and highly interactive analysis frame-by-frame, to a new era where automated visual analytics of very large datasets becomes the norm. In short, our goal is to ensure that the newly developed methods are used by industry and academic researchers in other areas, and turned into products for societal and economic benefit. To this end open source software, datasets, and demonstrators will be disseminated on the project website. The ubiquity of digital images and videos means that every UK citizen may potentially benefit from the Programme research in different ways. One example is smart audio-visual glasses, that can pay attention to a person talking by using their lip movements to mask out other ambient sounds. A second is an app that can answer visual questions (or retrieve matches) for text-queries over large scale audio-visual collections, such as a person's entire personal videos. A third is AI-guided medical screening, that can aid a minimally trained healthcare professional to perform medical scans.