Wessel Bruinsma ‡ wpb23@cam.ac.uk 1. Introduction A principled approach to prediction tasks is to choose a statistical model that explains the data. The choice of the model class is crucial and has to observe the bias-variance tradeoff, which motivates the need for principled approaches to selecting the best model class from a set of options. Whilst model selection can be done manually by trial and error, the process tends to consume considerable time and resources and be prone to human biases. Bayesian model selection (MacKay, 1992; Rasmussen and Ghahramani, 2001), treats the model class as a random variable and computes its posterior distribution. It offers a built-in complexity regulariser, commonly known as Bayesian Occams razor, which penalises models whose complexity is excessive or too modest.

In the last years, crowdsourcing is transforming the way classification training sets are obtained. Instead of relying on a single expert annotator, crowdsourcing shares the labelling effort among a large number of collaborators. For instance, this is being applied to the data acquired by the laureate Laser Interferometer Gravitational Waves Observatory (LIGO), in order to detect glitches which might hinder the identification of true gravitational-waves. The crowdsourcing scenario poses new challenging difficulties, as it deals with different opinions from a heterogeneous group of annotators with unknown degrees of expertise. Probabilistic methods, such as Gaussian Processes (GP), have proven successful in modeling this setting. However, GPs do not scale well to large data sets, which hampers their broad adoption in real practice (in particular at LIGO). This has led to the recent introduction of deep learning based crowdsourcing methods, which have become the state-of-the-art. However, the accurate uncertainty quantification of GPs has been partially sacrificed. This is an important aspect for astrophysicists in LIGO, since a glitch detection system should provide very accurate probability distributions of its predictions. In this work, we leverage the most popular sparse GP approximation to develop a novel GP based crowdsourcing method that factorizes into mini-batches. This makes it able to cope with previously-prohibitive data sets. The approach, which we refer to as Scalable Variational Gaussian Processes for Crowdsourcing (SVGPCR), brings back GP-based methods to the state-of-the-art, and excels at uncertainty quantification. SVGPCR is shown to outperform deep learning based methods and previous probabilistic approaches when applied to the LIGO data. Moreover, its behavior and main properties are carefully analyzed in a controlled experiment based on the MNIST data set.

More than 1,000 Google workers have signed a public letter calling on their employer to commit to an aggressive "company-wide climate plan" that includes canceling contracts with the fossil fuel industry and halting its donations to climate change deniers. The letter, which is addressed to Google's chief financial officer, Ruth Porat, also calls for zero emissions by 2030 and "zero collaboration with entities enabling the incarceration, surveillance, displacement or oppression of refugees or frontline communities". "We're excited to keep building momentum as tech workers join millions of people all over the world acting boldly for a livable future," said Sharon Campbell-Crow, a senior technical writer for Google, by email. "Marginalized communities have worked for climate justice for decades; Google needs to catch up and stop funding climate denial." The public campaign by Google workers follows similar efforts by employees of Amazon and Microsoft.

According to a recent report by IDC, digital transformation spending is expected to surpass $6 trillion dollars within the next four years, and it's believed that enterprises globally will spend more than $1 trillion on digital transformation before the end of 2019 alone. The report also notes that industries like process and discrete manufacturing and transportation will be some of the biggest spenders. These investments are fueling the growth of machine learning (ML) and the internet of things (IoT) to improve customer experiences and operational efficiency and accuracy. As companies have begun adopting digital transformations, there are a few things I'm looking forward to seeing more of in 2020. According to a Network World article, "IDC predicts that the collective sum of the world's data will grow from 33 zettabytes this year [2018] to 175ZB by 2025, for a compounded annual growth rate of 61%." (One zettabyte is equal to one trillion gigabytes.)

Learn how Microsoft Partner, eSmart Systems, empowers utility companies to stay ahead of power grid maintenance issues. Maintenance of electrical grids is not only time consuming and costly, but it can also be very dangerous. By developing a connected drone that uses AI and cognitive services from Microsoft Azure, utility companies can reduce blackouts and inspect power lines more safely. Learn how eSmart Systems' innovation with Microsoft is helping utility companies better protect their personnel and serve their communities. To learn about more areas where your agency can benefit from AI, read the Gartner report: https://aka.ms/whereyoushoulduseAI

Deep learning is a subset of artificial intelligence, in particular, the field of machine learning. Deep learning uses a multi-layered artificial neural network to carry out a range of tasks, from fraud detection to speech recognition or language translation. Deep learning differs from traditional machine learning systems in that it is capable of self-learning and improving as it analyses large data sets. A highly flexible system it has a number of applications in business. In this article, we explain exactly what deep learning is and explore the ways that it is already transforming businesses. Deep learning is a function of artificial intelligence. It is designed to replicate the way that the human brain processes data. It also re-creates the patterns found in the brain's decision-making process. Sometimes called deep neural networking or neural learning, it is part of the wider field of machine learning. It is powered by networks that can carry out unsupervised learning. This process uses algorithms to analyse raw data, extracting information and presenting it in a structured, useful model. Often it is also used to process unstructured or unlabeled data.

This paper considers the problem of Phase Identification in power distribution systems. In particular, it focuses on improving supervised learning accuracies by focusing on exploiting some of the problem's information theoretic properties. This focus, along with recent advances in Information Theoretic Machine Learning (ITML), helps us to create two new techniques. The first transforms a bound on information losses into a data selection technique. This is important because phase identification data labels are difficult to obtain in practice. The second interprets the properties of distribution systems in the terms of ITML. This allows us to obtain an improvement in the representation learned by any classifier applied to the problem. We tested these two techniques experimentally on real datasets and have found that they yield phenomenal performance in every case. In the most extreme case, they improve phase identification accuracy from $51.7\%$ to $97.3\%$. Furthermore, since many problems share the physical properties of phase identification exploited in this paper, the techniques can be applied to a wide range of similar problems.

Data-driven approaches are playing an increasingly significant role in chemical engineering. This session solicits submissions pertaining to application-driven methods and case studies demonstrating the use data and machine learning to infer correlations, develop models, as well as to improve processes/systems through data-driven optimization and control. Paper abstracts are public but to access Extended Abstracts, you must first purchase the conference proceedings. Paper abstracts are public but to access Extended Abstracts, you must first purchase the conference proceedings.

In a recent blog post, Google announced enhancements to a part of its Vision AI portfolio: AutoML Vision Edge, AutoML Video, and the Video Intelligence API. Each received updates to enhance their capabilities. Both AutoML Vision Edge and AutoML Video were both introduced earlier this year, in April, as a part of Google's AI Platform, while the Video Intelligence API introduction dates back a few years prior, with a public beta release in June 2017. We're constantly inspired by all the ways our customers use Google Cloud AI for image and video understanding--everything from eBay's use of image search to improve their shopping experience to AES leveraging AutoML Vision to accelerate a greener energy future and help make their employees safer. With AutoML Vision Edge, developers can train, build and deploy ML models at the edge.

Kongsberg Digital signed an agreement to digitalise the Nyhamna facility, a gas processing and export hub for Ormen Lange and other fields connected to the Polarled pipeline. A/S Norske Shell is entering the partnership as operator of Ormen Lange and on behalf of Gassco as the operator of Nyhamna. The value of the contract scope, with a digital twin, is approximately $11mn, with deliverables starting from Q4 2019. It will utilise the Kognifai Dynamic Digital Twin to create a virtual representation of the gas plant and its behavior – continuously updated with integrated information reflecting the status of the facility in real time. As the technical service provider at Nyhamna, Shell will be equipped with the ability to simulate scenarios and uncover new options for optimisation of its real-life counterpart.