Google's facial recognition video doorbell, the Nest Hello, is launching in the UK to challenge Amazon's Ring. Initially released in the US in March, Nest Hello is the first of Google's new home security product lineup to make it to the UK, ahead of its Nest Secure alarm system. Nest Hello is a £229 wifi-connected smart doorbell with a wide-angle camera that captures high definition HDR video with night vision after dusk. It will send alerts with pictures from the camera to users' phones and allows them to talk to visitors through the doorbell from anywhere with an internet connection. Users can also opt to have the doorbell say one of three quick phrases, such as "we'll be right there".

Skin cancer detection won't be turned over to machines anytime soon, but artificial intelligence detected skin cancer more accurately than a large group of international dermatologists in controlled testing, Agence France Presse reports. In an academic study and clinical trial published in Annals of Oncology, the study's lead author, Professor Holger A. Haenssle, of the University of Heidelberg Department of Dermatology, wrote, "Most dermatologists were outperformed by the CNN. Regardless of any physician's level of experience, they may benefit from assistance by a CNN's image classification." The study pitted 58 dermatologists from 17 countries against a deep learning convolutional neural network (CNN). Prior to the test, researchers from Germany, France, and the U.S. taught the CNN to differentiate benign skin lesions from dangerous melanomas. In the process, the team showed more than 100,000 images of correctly identified skin cancers to the neural network, which was designed with Google's Inception v4 CNN architecture.

Susceptibility of deep neural networks to adversarial attacks poses a major theoretical and practical challenge. All efforts to harden classifiers against such attacks have seen limited success. Two distinct categories of samples to which deep networks are vulnerable, "adversarial samples" and "fooling samples", have been tackled separately so far due to the difficulty posed when considered together. In this work, we show how one can address them both under one unified framework. We tie a discriminative model with a generative model, rendering the adversarial objective to entail a conflict. Our model has the form of a variational autoencoder, with a Gaussian mixture prior on the latent vector. Each mixture component of the prior distribution corresponds to one of the classes in the data. This enables us to perform selective classification, leading to the rejection of adversarial samples instead of misclassification. Our method inherently provides a way of learning a selective classifier in a semi-supervised scenario as well, which can resist adversarial attacks. We also show how one can reclassify the rejected adversarial samples.

Machine learning predictors are successfully deployed in applications ranging from disease diagnosis, to predicting credit scores, to image recognition. Even when the overall accuracy is high, the predictions often have systematic biases that harm specific subgroups, especially for subgroups that are minorities in the training data. We develop a rigorous framework of multiaccuracy auditing and post-processing to improve predictor accuracies across identifiable subgroups. Our algorithm, MultiaccuracyBoost, works in any setting where we have black-box access to a predictor and a relatively small set of labeled data for auditing. We prove guarantees on the convergence rate of the algorithm and show that it improves overall accuracy at each step. Importantly, if the initial model is accurate on an identifiable subgroup, then the post-processed model will be also. We demonstrate the effectiveness of this approach on diverse applications in image classification, finance, and population health. MultiaccuracyBoost can improve subpopulation accuracy (e.g. for `black women') even when the sensitive features (e.g. `race', `gender') are not known to the algorithm.

We present a computational framework for the semantic interpretation of symmetry in naturalistic scenes. Key features include a human-centred representation, and a declarative, explainable interpretation model supporting deep semantic question-answering founded on an integration of methods in knowledge representation and computer vision. In the backdrop of the visual arts, we showcase the framework's capability to generate human-centred, queryable, relational structures, also evaluating the framework with an empirical study on the human perception of visual symmetry. Our framework represents and is driven by the application of foundational Vision and KR methods in the psychological and social sciences.

Answer Set Programming (ASP) is a well-known problem solving approach based on nonmonotonic logic programs. HEX-programs extend ASP with external atoms for accessing arbitrary external information, which can introduce values that do not appear in the input program. In this work we consider inconsistent ASP- and HEX-programs, i.e., programs without answer sets. We study characterizations of inconsistency, introduce a novel notion for explaining inconsistencies in terms of input facts, analyze the complexity of reasoning tasks in context of inconsistency analysis, and present techniques for computing inconsistency reasons. This theoretical work is motivated by two concrete applications, which we also present. The first one is the new modeling technique of query answering over subprograms as a convenient alternative to the well-known saturation technique. The second application is a new evaluation algorithm for HEX-programs based on conflict-driven learning for programs with multiple components: while for certain program classes previous techniques suffer an evaluation bottleneck, the new approach shows significant, potentially exponential speedup in our experiments. Since well-known ASP extensions such as constraint ASP and DL-programs correspond to special cases of HEX, all presented results are interesting beyond the specific formalism.

We consider a particular instance of a common problem in recommender systems: using a database of book reviews to inform user-targeted recommendations. In our dataset, books are categorized into genres and sub-genres. To exploit this nested taxonomy, we use a hierarchical model that enables information pooling across across similar items at many levels within the genre hierarchy. The main challenge in deploying this model is computational: the data sizes are large, and fitting the model at scale using off-the-shelf maximum likelihood procedures is prohibitive. To get around this computational bottleneck, we extend a moment-based fitting procedure proposed for fitting single-level hierarchical models to the general case of arbitrarily deep hierarchies. This extension is an order of magnetite faster than standard maximum likelihood procedures. The fitting method can be deployed beyond recommender systems to general contexts with deeply-nested hierarchical generalized linear mixed models.

The forecasting and reconstruction of ocean and atmosphere dynamics from satellite observation time series are key challenges. While model-driven representations remain the classic approaches, data-driven representations become more and more appealing to benefit from available large-scale observation and simulation datasets. In this work we investigate the relevance of recently introduced bilinear residual neural network representations, which mimic numerical integration schemes such as Runge-Kutta, for the forecasting and assimilation of geophysical fields from satellite-derived remote sensing data. As a case-study, we consider satellite-derived Sea Surface Temperature time series off South Africa, which involves intense and complex upper ocean dynamics. Our numerical experiments demonstrate that the proposed patch-level neural-network-based representations outperform other data-driven models, including analog schemes, both in terms of forecasting and missing data interpolation performance with a relative gain up to 50\% for highly dynamic areas.

Deep learning systems have become ubiquitous in many aspects of our lives. Unfortunately, it has been shown that such systems are vulnerable to adversarial attacks, making them prone to potential unlawful and harmful uses. Designing deep neural networks that are robust to adversarial attacks is a fundamental step in making such systems safer and deployable in a broader variety of applications (e.g., autonomous driving), but more importantly is a necessary step to design novel and more advanced architectures built on new computational paradigms rather than marginally modifying existing ones. In this paper we introduce PeerNets, a novel family of convolutional networks alternating classical Euclidean convolutions with graph convolutions to harness information from a graph of peer samples. This results in a form of non-local forward propagation in the model, where latent features are conditioned on the global structure induced by the data graph, that is up to 3 more robust to a variety of white-and black-box adversarial attacks compared to conventional architectures with almost no drop in accuracy.

Reinforcement learning (RL) has advanced greatly in the past few years with the employment of effective deep neural networks (DNNs) on the policy networks. With the great effectiveness came serious vulnerability issues with DNNs that small adversarial perturbations on the input can change the output of the network. Several works have pointed out that learned agents with a DNN policy network can be manipulated against achieving the original task through a sequence of small perturbations on the input states. In this paper, we demonstrate furthermore that it is also possible to impose an arbitrary adversarial reward on the victim policy network through a sequence of attacks. Our method involves the latest adversarial attack technique, Adversarial Transformer Network (ATN), that learns to generate the attack and is easy to integrate into the policy network. As a result of our attack, the victim agent is misguided to optimise for the adversarial reward over time. Our results expose serious security threats for RL applications in safety-critical systems including drones, medical analysis, and self-driving cars.