Approximating a probability density in a tractable manner is a central task in Bayesian statistics. Variational Inference (VI) is a popular technique that achieves tractability by choosing a relatively simple variational family. Borrowing ideas from the classic boosting framework, recent approaches attempt to \emph{boost} VI by replacing the selection of a single density with a greedily constructed mixture of densities. In order to guarantee convergence, previous works impose stringent assumptions that require significant effort for practitioners. Specifically, they require a custom implementation of the greedy step (called the LMO) for every probabilistic model with respect to an unnatural variational family of truncated distributions. Our work fixes these issues with novel theoretical and algorithmic insights. On the theoretical side, we show that boosting VI satisfies a relaxed smoothness assumption which is sufficient for the convergence of the functional Frank-Wolfe (FW) algorithm. Furthermore, we rephrase the LMO problem and propose to maximize the Residual ELBO (RELBO) which replaces the standard ELBO optimization in VI. These theoretical enhancements allow for black box implementation of the boosting subroutine. Finally, we present a stopping criterion drawn from the duality gap in the classic FW analyses and exhaustive experiments to illustrate the usefulness of our theoretical and algorithmic contributions.

EPFL, Switzerland We present a system for the automatic differentiation of a higher-order functional array-processing language. The core functional language underlying this system simultaneously supports both sourceto-source automatic differentiation and global optimizations such as loop transformations. Thanks to this feature, we demonstrate how for some real-world machine learning and computer vision benchmarks, the system outperforms the state-of-the-art automatic differentiation tools. This investigation led him to see the importance of functional arguments and recursive functions in the field of symbolic computation. From Norvig [38, p248]. 1 INTRODUCTION Functional programming (FP) and automatic differentiation (AD) have been natural partners for sixty years, and major functional languages all have elegant automatic differentiation packages [6, 17, 29]. With the increasing importance of numerical engineering disciplines such as machine learning, speech processing, and computer vision, there has never been a greater need for systems which mitigate the tedious and error-prone process of manual coding of derivatives. However the popular packages (TensorFlow, CNTK) all implement clunky (E)DSLs in procedural languages such as Python and C . One reason is that the FP packages are slower than their imperative counterparts, by many orders of magnitude [48], because modern applications depend heavily on array processing, with vectors, matrices, and tensors as the canonical datatypes. In contrast, AD for FP has generally handled only scalar workloads efficiently [29]. Our key contribution in this paper is to take a recently introduced F# subset designed for efficient compilation of array-processing workloads, and to augment it with vector AD primitives, yielding a functional AD tool that is competitive with the best C/C and Fortran tools on many benchmarks, and considerably faster on others.

Non-intrusive load monitoring (NILM) refers to the technique of estimating the power demand of a single appliance from the combined demand of multiple appliances in a household measured by a single meter [1]. It is suggested in [2] that electricity consumption feedback that includes appliancespecific breakdown is more likely to promote electricity conservation for residential consumers. Electricity providers can have more detailed and in-depth understanding of their customers and provide better services. Thus, both electricity consumers and electricity providers can benefit from the information provided by accurate disaggregation of wholehome power demands. Comprehensive reviews of various NILM methods can be found in [3, 4].

The wide usage of Machine Learning (ML) has lead to research on the attack vectors and vulnerability of these systems. The defenses in this area are however still an open problem, and often lead to an arms race. We define a naive, secure classifier at test time and show that a Gaussian Process (GP) is an instance of this classifier given two assumptions: one concerns the distances in the training data, the other rejection at test time. Using these assumptions, we are able to show that a classifier is either secure, or generalizes and thus learns. Our analysis also points towards another factor influencing robustness, the curvature of the classifier. This connection is not unknown for linear models, but GP offer an ideal framework to study this relationship for nonlinear classifiers. We evaluate on five security and two computer vision datasets applying test and training time attacks and membership inference. We show that we only change which attacks are needed to succeed, instead of alleviating the threat. Only for membership inference, there is a setting in which attacks are unsuccessful (<10% increase in accuracy over random guess). Given these results, we define a classification scheme based on voting, ParGP. This allows us to decide how many points vote and how large the agreement on a class has to be. This ensures a classification output only in cases when there is evidence for a decision, where evidence is parametrized. We evaluate this scheme and obtain promising results.

Bayesian On-line Changepoint Detection is extended to on-line model selection and non-stationary spatio-temporal processes. We propose spatially structured Vector Autoregressions (VARs) for modelling the process between changepoints (CPs) and give an upper bound on the approximation error of such models. The resulting algorithm performs prediction, model selection and CP detection on-line. Its time complexity is linear and its space complexity constant, and thus it is two orders of magnitudes faster than its closest competitor. In addition, it outperforms the state of the art for multivariate data.

The robots are coming but while they might steal many of our jobs and cause significant social upheaval, it isn't all bad, delegates at the FutureScope conference in Dublin were told on Thursday. Speakers at the one-day event said there is no doubt that automation is changing the way we work and that more must be done to ensure we manage the changes effectively. Siobhan O'Shea, client services director at listed technology recruitment firm CPL, said the impact of robots in the workplace was already beginning to be felt. The rise of automation will lead "not to mass unemployment, but mass redeployment" of workers," she said. Ms O'Shea said education is a key area in which preparation for future changes needs to be made. She noted that the number of teachers in Ireland with qualifications in biology currently outnumbers those with qualifications in physics by three to one and that this needs to change. "Ireland has one of the lowest rates in Europe for lifelong learning so it presents a systemic challenge for us," she said. "With advances in technology growing what people learn at college or university now will be out of date within two years," Ms O'Shea added. Anthony Behan, industry lead in IBM's Watson IoT division said while there will be job losses, there will also be opportunities, many of which we've yet to imagine. "We have enormous amounts of jobs being created in new areas due to technological innovation, it isn't just about losing them, Mr Behan said.

Experts have raised ethical issues surrounding this technology before. The Malicious AI report focused on fake videos to make people believe false information, and could jeopardize political security. The paper doesn't address these concerns too much. But it did say that pushing the limits of this technology and democratising "calls for additional care in ensuring verifiable video authenticity, e.g., through invisible watermarking." Justus Thies, a coauthor of the paper and a postdoctoral researcher at the Technical University of Munich, in Germany, told The Register that he recognized the potential dangers of using AI to manipulate fake videos.

Mount Sinai Health System on Friday announced an exclusive multiyear license and partnership with RenalytixAI, an artificial-intelligence startup with offices in New York and the United Kingdom. The goal is to reduce the $98 billion in preventable kidney disease and dialysis costs by predicting which patients are at the greatest risk of advanced kidney disease and taking steps to treat them early on. The venture will draw from the more than 3 million electronic health records in Mount Sinai's system, plus an additional 43,000 patient records in Mount Sinai's BioMe BioBank, part of the Charles Bronfman Institute for Personalized Medicine. The bank collects DNA and blood serum from a diverse patient base to identify biomarkers, substances that can indicate disease, infection or environmental exposure. All data will be de-identified to protect patient privacy "We can look at relationships that we could never look at before," said RenalytixAI CEO James McCullough.

Researchers have taken another step towards realistic, synthesized video. The team, made up of scientists in Germany, France, the UK and the US, used AI to transfer the head poses, facial expressions, eye motions and blinks of a person in one video onto another entirely different person in a separate video. The researchers say it's the first time a method has transferred these types of movements between videos and the result is a series of clips that look incredibly realistic. The neural network created by the researchers only needs a few minutes of the target video for training and it can then translate the head, facial and eye movements of the source to the target. It can even manipulate some background shadows when they're present.

The government has unveiled a new £30 million fund to help get driverless cars on the road. Chancellor Philip Hammond vowed to bring fully-autonomous vehicles to the UK by 2021 in his autumn budget last year. The latest multi-million round of funding is designed to speed-up the roll-out of autonomous vehicles by supporting technology and automotive companies developing driverless systems. Companies will have to bid for the funding, Business Minister Richard Harrington announced today. Of the £30 million total, the government has set aside £5 million to be awarded specifically to projects building cars that can park themselves.