From traffic control and waste management to biometric surveillance systems and predictive policing models, the potential uses of artificial intelligence (AI) in cities are incredibly diverse, and could impact every aspect of urban life. In response to the increasing deployment of AI in cities – and the general lack of authority that municipal governments have to challenge central government decisions or legislate themselves – London, Barcelona and Amsterdam launched the Global Observatory on Urban AI in June 2021. The initiative aims to monitor AI deployment trends and promote its ethical use, and is part of the wider Cities Coalition for Digital Rights (CC4DR), which was set up in November 2018 by Amsterdam, Barcelona and New York to promote and defend digital rights. It now has more than 50 cities participating worldwide. Apart from city participants, the Observatory is also being run in partnership with UN-Habitat, a United Nations initiative to improve the quality of life in urban areas, and research group CIDOB-Barcelona Centre for International Affairs.

Artificial Intelligence (AI) is progressively being ingrained in the fabric of business and is being broadly applied across a variety of application use cases. However, not all industries are at the same level of AI adoption: the information technology and telecommunications industry are the most advanced, with the automobile sector trailing closely behind. AI may be both a boon and a disaster for cybersecurity since it is a general-purpose, dual-use technology. This is supported by the fact that AI is employed both as a sword (i.e. to enable malicious assaults) and as a shield (i.e. to defend against them) (to counter cybersecurity risks). The cyberattack surface in today's business environments is enormous, and it's just becoming bigger. AI and machine learning are becoming increasingly important in cybersecurity because they can quickly analyze billions of data sets and hunt down a wide range of cyber risks, from malware to shady conduct that might lead to a phishing assault.

This announcement accelerates Optellum's market entry, building on its FDA clearance earlier this year and deployments underway at hospitals in the USA and ongoing clinical trials in the United Kingdom. It identifies and tracks at-risk patients and assigns a Lung Cancer Prediction score to lung nodules: small lesions, frequently detected in chest Computed Tomography (CT) scans that may or may not be cancerous. The Optellum AI will be used to drive accurate early diagnosis and optimal treatment decisions with the aim of treating patients earlier, potentially at a pre-cancerous stage, increasing survival rates.

Accurate and interpretable prediction of survey response rates is important from an operational standpoint. The US Census Bureau's well-known ROAM application uses principled statistical models trained on the US Census Planning Database data to identify hard-to-survey areas. An earlier crowdsourcing competition revealed that an ensemble of regression trees led to the best performance in predicting survey response rates; however, the corresponding models could not be adopted for the intended application due to limited interpretability. In this paper, we present new interpretable statistical methods to predict, with high accuracy, response rates in surveys. We study sparse nonparametric additive models with pairwise interactions via $\ell_0$-regularization, as well as hierarchically structured variants that provide enhanced interpretability. Despite strong methodological underpinnings, such models can be computationally challenging -- we present new scalable algorithms for learning these models. We also establish novel non-asymptotic error bounds for the proposed estimators. Experiments based on the US Census Planning Database demonstrate that our methods lead to high-quality predictive models that permit actionable interpretability for different segments of the population. Interestingly, our methods provide significant gains in interpretability without losing in predictive performance to state-of-the-art black-box machine learning methods based on gradient boosting and feedforward neural networks. Our code implementation in python is available at https://github.com/ShibalIbrahim/Additive-Models-with-Structured-Interactions.

Many autonomous systems, such as driverless taxis, perform safety critical functions. Autonomous systems employ artificial intelligence (AI) techniques, specifically for the environment perception. Engineers cannot completely test or formally verify AI-based autonomous systems. The accuracy of AI-based systems depends on the quality of training data. Thus, novelty detection - identifying data that differ in some respect from the data used for training - becomes a safety measure for system development and operation. In this paper, we propose a new architecture for autoencoder-based semantic novelty detection with two innovations: architectural guidelines for a semantic autoencoder topology and a semantic error calculation as novelty criteria. We demonstrate that such a semantic novelty detection outperforms autoencoder-based novelty detection approaches known from literature by minimizing false negatives.

You love your retirement-age parents. You want what's best for them. But let me ask you this: Do you know where your parents are right now? We live in a scary world, and today's parents require constant supervision. When it comes to the safety of your aging mom and dad, you can never be too vigilant.

Information manipulation has been around since Chinese general Sun Tzu wrote "The Art of War" in 550 BC. The Russians routinely use disinformation tactics to destabilize democracies. Events like the 2020 U.S. elections or COVID-19 vaccinations highlight how political opponents and rogue nations actively practice disinformation campaigns to undermine confidence in governments and science, sowing fear and distrust. The disinformation machine is said to cost the global economy $78 billion yearly. The good news is that we're getting better at detecting deep fakes.

The industry thrives on its futuristic image, worships boy-CEOs and renders the past obsolete at a frightening pace. Even in the eight years I've sat on the Guardian's technology desk, the field I cover is frequently unrecognisable from what it was when I started – a world where self-driving cars were just around the corner, where virtual reality was an impressive technology that had failed to catch on with normal people, and where the world was starting to tire of the like-clockwork appearance of a new iPhone every 12 months. Well, fine, but some things really have changed in that time. Just before I started at the paper, the Guardian broke the news that the NSA had been spying on Americans – and the rest of the world – through the tech sector, with more revelations to come thanks to the whistleblowing efforts of Edward Snowden. It was the first sign that the lustre had started to come off the sector, an inkling of what was to follow a few years later as the "techlash" saw first Facebook, then the rest of the industry, fall from grace.

The past decade has seen a substantial rise in the amount of mis- and disinformation online, from targeted disinformation campaigns to influence politics, to the unintentional spreading of misinformation about public health. This development has spurred research in the area of automatic fact checking, from approaches to detect check-worthy claims and determining the stance of tweets towards claims, to methods to determine the veracity of claims given evidence documents. These automatic methods are often content-based, using natural language processing methods, which in turn utilise deep neural networks to learn higher-order features from text in order to make predictions. As deep neural networks are black-box models, their inner workings cannot be easily explained. At the same time, it is desirable to explain how they arrive at certain decisions, especially if they are to be used for decision making. While this has been known for some time, the issues this raises have been exacerbated by models increasing in size, and by EU legislation requiring models to be used for decision making to provide explanations, and, very recently, by legislation requiring online platforms operating in the EU to provide transparent reporting on their services. Despite this, current solutions for explainability are still lacking in the area of fact checking. This thesis presents my research on automatic fact checking, including claim check-worthiness detection, stance detection and veracity prediction. Its contributions go beyond fact checking, with the thesis proposing more general machine learning solutions for natural language processing in the area of learning with limited labelled data. Finally, the thesis presents some first solutions for explainable fact checking.

This paper presents an innovative deep learning pipeline which estimates the relative pose of a spacecraft by incorporating the temporal information from a rendezvous sequence. It leverages the performance of long short-term memory (LSTM) units in modelling sequences of data for the processing of features extracted by a convolutional neural network (CNN) backbone. Three distinct training strategies, which follow a coarse-to-fine funnelled approach, are combined to facilitate feature learning and improve end-to-end pose estimation by regression. The capability of CNNs to autonomously ascertain feature representations from images is exploited to fuse thermal infrared data with red-green-blue (RGB) inputs, thus mitigating the effects of artefacts from imaging space objects in the visible wavelength. Each contribution of the proposed framework, dubbed ChiNet, is demonstrated on a synthetic dataset, and the complete pipeline is validated on experimental data.