AI that can tell pre-cancerous growths from harmless moles - great. As a technology, AI is unprecedented, powerful and deeply pervasive: from voice recognition to self-driving cars to medical diagnosis, it is swiftly weaving its way into our lives at work, home, and everywhere in between. Yet most of us know very little about it. It's easy to say that fear of the unknown is fruitless, but how can we understand and leverage AI to create the best possible society without succumbing to fear, doomsaying, or prophecies of a fate worse than war? In 2016, the news broke that US risk assessment algorithms - used by courtrooms throughout the country to decide the fates and freedoms of those on trial - are racially biased, frequently sentencing Caucasians more leniently than African Americans despite no difference in the type of crime committed.

Artificial Intelligence in Construction is here and its influence will continue to grow. I'm interested to find out how it will effect multiple aspects of the construction process from the manufacturing of products itself, to the distribution and the installation. How will its introduction be slowed by costs, reluctance to change and most importantly how will it influence the role of the workforce in the construction industry. Artificial Intelligence has always been a fascination for me and for many I'm sure. From Terminator and Robocop to i-Robot and Westworld, the prospect of a world populated by robots is common in Hollywood and all forms of visual entertainment. In the UK we have already had the pleasure of self-service checkouts for almost a decade.

Back in April, NASA and the European Space Agency (ESA) agreed to figure out whether it's possible to bring Martian soil samples back to Earth. Now, the ESA has granted Airbus a $5.2 million contract to design a concept for a rover that can collect those samples on the red planet. The space agency has chosen Airbus' team in Stevenage, England, because they're already building the ExoMars rover that's scheduled to head to our neighboring planet in 2021. Unlike ExoMars and all its scientific instruments, the fetch rover's lone task is to find and collect the canisters of samples NASA's Mars 2020 rover prepares and leaves behind. However, that doesn't mean that designing the vehicle would be easy.

The traditional banking business is undergoing an exciting period of disruption. Big data, blockchain, an eruption of new competitors of all shapes and sizes … With so much that is new, we run the risk of overlooking how artificial intelligence is already beginning to change the core of the financial business. Its impact is clearly manifest in five areas: Credit scoring (or creditworthiness or rating), market research, personal assistants, asset management, and fraud detection. Startups like Kensho, recently acquired for $550 million, and Dataminr use artificial intelligence algorithms to improve the management of financial assets. Dataminr is specifically focused on identifying patterns and indexes via social networks, whereas Kensho stands out for its ability to establish correlations between news – from Brexit to natural catastrophes – and the markets.

Fortnite is the biggest game in the world, and has turned millions of its young players into obsessives. But that obsession is being exploited by scammers who try and steal information and money from the children who play it. A whole host of scams are being used across the internet to try and trick children into giving up the most personal and sensitive information of themselves and their parents. And it is being done by offering them free money and winning them over by chatting to them through headsets. Fortnite Battle Royale was released around a year ago, as a free-to-download game that is now available for just about every computing platform. It sees players attempt to stay alive for as long as they can, in games that last around 20 minutes, either by killing other players or successfully hiding from them.

Structural Causal Models are widely used in causal modelling, but how they relate to other modelling tools is poorly understood. In this paper we provide a novel perspective on the relationship between Ordinary Differential Equations and Structural Causal Models. We show how, under certain conditions, the asymptotic behaviour of an Ordinary Differential Equation under non-constant interventions can be modelled using Dynamic Structural Causal Models. In contrast to earlier work, we study not only the effect of interventions on equilibrium states; rather, we model asymptotic behaviour that is dynamic under interventions that vary in time, and include as a special case the study of static equilibria.

Interpretability of deep neural networks is a recently emerging area of machine learning research targeting a better understanding of how models perform feature selection and derive their classification decisions. In this paper, two neural network architectures are trained on spectrogram and raw waveform data for audio classification tasks on a newly created audio dataset and layer-wise relevance propagation (LRP), a previously proposed interpretability method, is applied to investigate the models' feature selection and decision making. It is demonstrated that the networks are highly reliant on feature marked as relevant by LRP through systematic manipulation of the input data. Our results show that by making deep audio classifiers interpretable, one can analyze and compare the properties and strategies of different models beyond classification accuracy, which potentially opens up new ways for model improvements.

This paper describes how to apply self-attention with relative positional encodings to the task of relation extraction. We propose to use the self-attention encoder layer together with an additional position-aware attention layer that takes into account positions of the query and the object in the sentence. The self-attention encoder also uses a custom implementation of relative positional encodings which allow each word in the sentence to take into account its left and right context. The evaluation of the model is done on the TACRED dataset. The proposed model relies only on attention (no recurrent or convolutional layers are used), while improving performance w.r.t. the previous state of the art.

Given a malfunctioning system, sequential diagnosis aims at identifying the root cause of the failure in terms of abnormally behaving system components. As initial system observations usually do not suffice to deterministically pin down just one explanation of the system's misbehavior, additional system measurements can help to differentiate between possible explanations. The goal is to restrict the space of explanations until there is only one (highly probable) explanation left. To achieve this with a minimal-cost set of measurements, various (active learning) heuristics for selecting the best next measurement have been proposed. We report preliminary results of extensive ongoing experiments with a set of selection heuristics on real-world diagnosis cases. In particular, we try to answer questions such as "Is some heuristic always superior to all others?", "On which factors does the (relative) performance of the particular heuristics depend?" or "Under which circumstances should I use which heuristic?"

One of the fundamental challenges in supervised learning for multimodal image registration is the lack of ground-truth for voxel-level spatial correspondence. This work describes a method to infer voxel-level transformation from higher-level correspondence information contained in anatomical labels. We argue that such labels are more reliable and practical to obtain for reference sets of image pairs than voxel-level correspondence. Typical anatomical labels of interest may include solid organs, vessels, ducts, structure boundaries and other subject-specific ad hoc landmarks. The proposed end-to-end convolutional neural network approach aims to predict displacement fields to align multiple labelled corresponding structures for individual image pairs during the training, while only unlabelled image pairs are used as the network input for inference. We highlight the versatility of the proposed strategy, for training, utilising diverse types of anatomical labels, which need not to be identifiable over all training image pairs. At inference, the resulting 3D deformable image registration algorithm runs in real-time and is fully-automated without requiring any anatomical labels or initialisation. Several network architecture variants are compared for registering T2-weighted magnetic resonance images and 3D transrectal ultrasound images from prostate cancer patients. A median target registration error of 3.6 mm on landmark centroids and a median Dice of 0.87 on prostate glands are achieved from cross-validation experiments, in which 108 pairs of multimodal images from 76 patients were tested with high-quality anatomical labels.