A core novelty of Alpha Zero is the interleaving of tree search and deep learning, which has proven very successful in board games like Chess, Shogi and Go. These games have a discrete action space. However, many real-world reinforcement learning domains have continuous action spaces, for example in robotic control, navigation and self-driving cars. This paper presents the necessary theoretical extensions of Alpha Zero to deal with continuous action space. We also provide some preliminary experiments on the Pendulum swing-up task, empirically showing the feasibility of our approach. Thereby, this work provides a first step towards the application of iterated search and learning in domains with a continuous action space.

We compare the fast training and decoding speed of RETURNN of attention models for translation, due to fast CUDA LSTM kernels, and a fast pure TensorFlow beam search decoder. We show that a layer-wise pretraining scheme for recurrent attention models gives over 1% BLEU improvement absolute and it allows to train deeper recurrent encoder networks. Promising preliminary results on max. expected BLEU training are presented. We are able to train state-of-the-art models for translation and end-to-end models for speech recognition and show results on WMT 2017 and Switchboard. The flexibility of RETURNN allows a fast research feedback loop to experiment with alternative architectures, and its generality allows to use it on a wide range of applications.

In this talk, Richard Garris, Principal Architect at Databricks will explain how various ML algorithms are parallelized in Apache Spark. Andrew Ng calls the algorithms the "rocket ship" and the data "the fuel that you feed machine learning" to build deep learning applications. We will start with an understanding of machine learning pipelines built using single machine algorithms including Pandas, scikit-learn, and R. Then we will discuss how Apache Spark MLlib can be used to parallelize your machine learning pipeline with Linear Regression and Random Forest. Lastly, we will discuss ways to parallelize single machine algorithms in Spark by broadcasting the data and then performing distributed feature selection, model creation or hyperparameter tuning. Bio: Richard Garris is a Principal Solutions Architect at Databricks focused on helping clients with their Advanced Analytics initiatives using Apache Spark and MLlib.

At least half of the world's population cannot obtain essential health services. But low-cost, easy-to-use technologies powered by Artificial Intelligence (AI) promise to deliver quality and affordable health care to the people who need it most, no matter how hard to reach. At the AI for Good Global Summit last week, entrepreneurs, AI experts, academics and UN representatives described many AI technologies for health, allowing for the early detection of various pathologies such as osteoarthritis, diabetic retinopathy, child malnutrition, snakebites and others. These technologies don't place a heavy burden on doctors, and can lead to prompt diagnosis and effective treatment. They agreed that AI can add tremendous value in developing countries where there is a low density of physicians.

From the sprawling open world Western Red Dead Redemption 2, to the co-op dinosaur hunting Monster Hunter World; from major remakes like Shadow of the Colossus to brand new IPs like Days Gone; from charming JRPGs like Ni No Kuni II, to terrifying horror games like Agony, 2018 is shaping up to be an incredible year for games and gamers. The following list is long but not exhaustive. It doesn't cover games that haven't been announced (like the inevitable Call of Duty from Treyarch) or that don't have titles yet (like the Fire Emblem game coming to Switch.) I've also left mobile games out almost entirely. These probably deserve their own list. Likewise, PC games release in such abundance (thousands on Steam each month!) that many indie PC titles have been left out. If I've missed anything you think ought to be included, please let me know on Twitter or Facebook. Now, without further ado, let's get to the games. The first proper fighting game in the Dragon Ball Z series, Dragon Ball FighterZ brings 3 vs 3 matches similar to Marvel vs Capcom with the ability to swap out characters during a fight. The first current-gen Monster Hunter entry, this also marks the series' biggest world wide release. The franchise is much more popular in Japan, but hopefully break into the West with Monster Hunter World's giant, seamless environments and online co-op. Whether or not you played the original Shadow of the Colossus, this remake (not remaster) looks absolutely gorgeous.

London – May 23, 2018 – DataRobot, a pioneer in automated machine learning, announced today the release of a new version of the DataRobot Cloud platform for the European Union (EU). The DataRobot EU Cloud, available on the Amazon Web Services (AWS) EU (Ireland) Region, is able to support the unique needs of EU customers, whose data is tightly regulated, as well as help enterprises conform to the upcoming General Data Protection Regulation (GDPR) requirements. With growing sensitivities over data privacy and breaches, the need for organisations to keep and process customer data at the point of origin is vital. With the introduction of GDPR and its 25 May 2018 enforcement date, organisations are also required to develop a strategy to prove compliance or face hefty fines. The DataRobot EU Cloud ensures European organisations adhere to enhanced data sovereignty requirements without sacrificing any enterprise-grade capabilities.

The Government has set out an ambition for data and artificial intelligence to help diagnose cancer at a much earlier stage for at least 50,000 people per year within the next 15 years. Prime Minister Theresa May (pictured) outlined this as a step in the AI and Data Grand Challenge – one of four in the Government's Industrial Strategy – in a speech yesterday on national ambitions in science and technology. It had already identified diagnosing medical conditions as the first challenge for the technology, and May's speech has now set the more specific target for cancer. She said that combined with treatment from the NHS it could mean that 22,000 fewer people will die within five years of their diagnosis compared to today. "We will work with industry and the medical research community to announce specific ambitions in a range of other disease areas over the coming weeks and months," she said.

With artificial intelligence integrated into building systems and IoT devices, buildings become more than their brick-and-mortar shells. They begin to operate in new ways, creating personalized experiences for their occupants, and providing energy and cost savings for their owners. In the report'Building intelligence into buildings', the Institute of Business Value (IBV) explores the potential of buildings that think for themselves. Below are some of the key learnings and action points. What if buildings owners could see exactly how their building is being used at any given time?

The Royal College Hospital London (UCLH) plans to use artificial intelligence to support clinical decision-making, replacing doctors and nurses in certain situations. The hospital will work alongside The Alan Turing Institute, a government-funded data science research hub, to look at ways to make NHS services quicker, safer and more efficient, announcing a three-year partnership. One area they will focus on is A&E, which is seen as a barometer of how the rest of the hospital and the wider system is working. It is hoped the technology can help the hospital achieve the national average waiting times of four hours, which it is currently not meeting. "Imagine a scenario where patients present to A&E with abdominal pain. Our standard response is to check bloods, order X-rays or scans and in probably about 80% of cases, discharge for home management," said Professor Marcel Levi, UCLH chief executive.

AI and automation is changing the business environment across industries, delivering new opportunities through intelligent, automated products. Some companies are ahead of the curve, and others are stagnating in their adoption of the tech. Board members and decision-makers are increasingly aware of the benefits of AI and automation, but the question should always remain: 'Is it right for my business? How does it solve a problem?'. With the general rise of this technology into business operations also comes challenges, dangers and potential risks to the human workforce.