Prowler.io CEO Vishal Chatrath and I are talking about artificial intelligence in a room called Maria. The meeting room, which forms part of Prowler's bright and buzzing Hills Road offices, doesn't have consciousness as far as I'm aware, but is named in honour of one of the main characters in Metropolis, a 1927 film which depicts an early vision of the rise of the machines. In it, Maria sees her likeness transferred to a robot, which leads an uprising to destroy the titular Metropolis. It's one of several nods to sci-fi at Prowler HQ – the firm's boardroom is called Skynet – but the company, which is being tipped by many as the Next Big Thing to emerge from the Cambridge cluster, has its sights set firmly on using AI to solve real-world problems. Vishal explains that these problems could be, well, just about anything.

What do queues, zoos, and machine learning have in common? Marwell Zoo is building better beds for their animals with IoT and machine learning. Using Watson IoT Platform, the park's keepers are working in tandem with IBM, designing a better way to reduce energy consumption. Can machine learning be used to create better conditions for animals at Marwell Zoo? The question first emerged after Andy Stanford-Clark, CTO for IBM UK & Ireland, presented an introduction to Internet of Things at a UK Chamber of Commerce conference in February 2017.

Current domain-independent, classical planners require symbolic models of the problem domain and instance as input, resulting in a knowledge acquisition bottleneck. Meanwhile, although deep learning has achieved significant success in many fields, the knowledge is encoded in a subsymbolic representation which is incompatible with symbolic systems such as planners. We propose LatPlan, an unsupervised architecture combining deep learning and classical planning. Given only an unlabeled set of image pairs showing a subset of transitions allowed in the environment (training inputs), and a pair of images representing the initial and the goal states (planning inputs), LatPlan finds a plan to the goal state in a symbolic latent space and returns a visualized plan execution. The contribution of this paper is twofold: (1) State Autoencoder, which finds a propositional state representation of the environment using a Variational Autoencoder. It generates a discrete latent vector from the images, based on which a PDDL model can be constructed and then solved by an off-the-shelf planner. (2) Action Autoencoder / Discriminator, a neural architecture which jointly finds the action symbols and the implicit action models (preconditions/effects), and provides a successor function for the implicit graph search. We evaluate LatPlan using image-based versions of 3 planning domains: 8-puzzle, Towers of Hanoi and LightsOut.

Generative Adversarial Networks (GANs) have shown remarkable success as a framework for training models to produce realistic-looking data. In this work, we propose a Recurrent GAN (RGAN) and Recurrent Conditional GAN (RCGAN) to produce realistic real-valued multidimensional time series, with an emphasis on their application to medical data. RGANs make use of recurrent neural networks (RNNs) in the generator and the discriminator. In the case of RCGANs, both of these RNNs are conditioned on auxiliary information. We demonstrate our models in a set of toy datasets, where we show visually and quantitatively (using sample likelihood and maximum mean discrepancy) that they can successfully generate realistic time-series. We also describe novel evaluation methods for GANs, where we generate a synthetic labelled training dataset, and evaluate on a real test set the performance of a model trained on the synthetic data, and vice-versa. We illustrate with these metrics that RCGANs can generate time-series data useful for supervised training, with only minor degradation in performance on real test data. This is demonstrated on digit classification from'serialised' MNIST and by training an early warning system on a medical dataset of 17,000 patients from an intensive care unit. We further discuss and analyse the privacy concerns that may arise when using RCGANs to generate realistic synthetic medical time series data, and demonstrate results from differentially private training of the RCGAN.

We introduce neural networks for end-to-end differentiable proving of queries to knowledge bases by operating on dense vector representations of symbols. These neural networks are constructed recursively by taking inspiration from the backward chaining algorithm as used in Prolog. Specifically, we replace symbolic unification with a differentiable computation on vector representations of symbols using a radial basis function kernel, thereby combining symbolic reasoning with learning subsymbolic vector representations. By using gradient descent, the resulting neural network can be trained to infer facts from a given incomplete knowledge base. It learns to (i) place representations of similar symbols in close proximity in a vector space, (ii) make use of such similarities to prove queries, (iii) induce logical rules, and (iv) use provided and induced logical rules for multi-hop reasoning. We demonstrate that this architecture outperforms ComplEx, a state-of-the-art neural link prediction model, on three out of four benchmark knowledge bases while at the same time inducing interpretable function-free first-order logic rules.

They enable new powerful analytics using built-in graph algorithms and the PGQL language. Since 2015 Oracle offers a graph database with Oracle Spatial and Graph. The graphs can be stored in Oracle NoSQL, Apache HBase or an Oracle Database. The analytics are performed by an in-memory engine for optimal performance. In this paper we explain the fundamentals of property graph databases and highlight use cases where property graph implementations are superior to relational technologies.

The future is always coming and yet it never quite seems to get here. Once upon a time, we were promised jetpacks. Likewise with those robot assistants that were meant to relieve 1950s housewives from the drudgery of cleaning. And then we have AI, which only started generating a buzz recently, but since then that swell of anticipation has dissipated somewhat. Is it destined to go the way of the jetpack and the robot home help or is it an actual thing that's going to make a profound difference to the way you approach marketing and to every other domain that AI threatens to disrupt?

AI is the buzzword on everyone's lips these days. But how might musicians respond to themes of machine intelligence? We're calling this year's theme "The Hacked Mind." Inspired by AI and machine learning, we're inviting artists to respond in the latest edition of our MusicMakers Hacklab hosted with CTM Festival in Berlin. In that collaborative environment, participants will have a chance to answer these questions however they like.

Machine learning helps make decisions by analyzing data and can be used in many different areas, including music choice and facial recognition. Yandex, one of Russia's leading tech companies, has made its advanced machine learning algorithm, CatBoost, available free of charge for developers around the globe. "This is the first Russian machine learning technology that's an open source," said Mikhail Bilenko, Yandex's head of machine intelligence and research. What do cats have to do with this?

Artificial intelligence is a hot topic in the tech industry with China -- alongside the U.S. -- emerging as a key market for talent, innovation and companies. While many AI firms head to the U.S. or set up operations Stateside, China is a harder market to crack for overseas companies. That's where a new fund wants to help. China-based accelerator firm COMB is launching a €65 million ($77 million) fund aimed at helping promising international AI firms enter the Middle Kingdom. The fund, which was announced at Slush in Helsinki this week, is run by COMB and the Beijing Institute of Collaborative Innovation (BICI).