Complex artifacts are, of necessity, constructed by assembling simpler components. Software systems use libraries of reusable components, and often access multiple remote services. In this paper, we consider systems that can be formalized as solving the model expansion task for some class of finite structures. A wide range of problem solving and query answering systems are so accounted for. We present a method for automatically generating a solver for a complex system from a declarative definition of that system in terms of simpler modules, together with solvers for those modules. The work is motivated primarily by "knowledge-intensive" computing contexts, where the individual modules are defined in (possibly different) declarative languages, such as logical theories or logic programs, but can be applied anywhere the model expansion formalization can. The Algebra of Modular Systems (AMS) [48, 49], provides a way to define a complex module in terms of a collection of other modules, in purely semantic terms. Formally, each module in this algebra represents a class of structures, and a "solver" for the module solves the model expansion task for that class. That is, a solver for module M takes as input a structure A for a part of the vocabulary of M, and returns either a set of expansions of A that are in M, or the empty set.

The ability of the Generative Adversarial Networks (GANs) framework to learn generative models mapping from simple latent distributions to arbitrarily complex data distributions has been demonstrated empirically, with compelling results showing that the latent space of such generators captures semantic variation in the data distribution. Intuitively, models trained to predict these semantic latent representations given data may serve as useful feature representations for auxiliary problems where semantics are relevant. However, in their existing form, GANs have no means of learning the inverse mapping -- projecting data back into the latent space. We propose Bidirectional Generative Adversarial Networks (BiGANs) as a means of learning this inverse mapping, and demonstrate that the resulting learned feature representation is useful for auxiliary supervised discrimination tasks, competitive with contemporary approaches to unsupervised and self-supervised feature learning.

Developments in semantic web technologies have promoted ontological encoding of knowledge from diverse domains. However, modelling many practical domains requires more expressive representations schemes than what the standard description logics(DLs) support. We extend the DL SROIQ with constraint networks and grounded circumscription. Applications of constraint modelling include embedding ontologies with temporal or spatial information, while grounded circumscription allows defeasible inference and closed world reasoning. This paper overcomes restrictions on existing constraint modelling approaches by introducing expressive constructs. Grounded circumscription allows concept and role minimization and is decidable for DL. We provide a general and intuitive algorithm for the framework of grounded circumscription that can be applied to a whole range of logics. We present the resulting logic: GC-SROIQ(C), and describe a tableau decision procedure for it.

In a strong indication of the links in tech between China and Silicon Valley, social network giant Tencent has acquired a 5 percent stake in U.S. electric car maker Tesla for $1.8 billion. The rationale behind the deal is potential collaboration on automated ride-sharing and delivery services as well as related information, entertainment and e-commerce content. Alibaba-backed search engine startup Quixey in Silicon Valley has failed with strategic misalignment, cultural differences and disagreements over a commercial contract to blame. Quixey was one of Alibaba's first bets in Silicon Valley tech startups, and the Chinese conglomerate led financing of well more than $100 million in Quixey, starting from 2013 and ending with a $30 million loan and $10 million in cash last summer. It's a high-profile fail for Alibaba and its U.S. investment strategy.

Up to around 30% of existing UK jobs could face automation over the next 15 years, but new AI-related technologies will also boost productivity and generate additional jobs elsewhere in the economy, according to new analysis by PwC in its latest UK Economic Outlook report. This involves looking in detail at the task composition of jobs in different industry sectors and occupations, using machine learning techniques to model the potential impact of AI in the future based on OECD data. The study estimates that the UK (30%) has a lower proportion of existing jobs at potential high risk of automation than the US (38%) and Germany (35%), but more than Japan (21%). PwC's analysis finds the likely impact of automation varies significantly across industry sectors: transportation and storage (56%), manufacturing (46%) and wholesale and retail trade (44%) have the highest proportion of jobs facing potential high risks of automation among the larger sectors. Education and health and social work are estimated to face the lowest risks of automation given the relatively high proportion of tasks that are hard to automate (see table below).

While brands have scrambled to launch Facebook Messenger chatbots since the social media behemoth opened up the channel for development last year, the early results haven't been particularly promising. Facebook is seeing a 70% failure rate among those 35,000 or so bots when it comes to understanding user requests. To combat this poor performance, Facebook is making some changes to Messenger, including adding a persistent menu that will allow users to choose from a number of requests or statements instead of using natural language and risking stumping the bot entirely. There's no question that AI will play a huge role in the future of retail, but in these early days of chatbots and virtual assistants, how do you reap the benefits while avoiding the pitfalls of this emerging technology? We caught up with Linc engineer Alessandro Sanchez to talk about the potential weaknesses in current chatbots and how smart brands are creating a chatbot experience that beats the odds and delivers great service.

Sign-up to download the forthcoming report: "Artificial Intelligence: Teaching Machines to Think Like People," by Jack Clark. This interview is one in a series of interviews that will be featured in the report. A better understanding of the reasons why neurons spike could lead to smart AI systems that can store more information more efficiently, according to Geoff Hinton, who is often referred to as the "godfather" of deep learning. Geoff Hinton is an emeritus distinguished professor at the University of Toronto and an engineering fellow at Google. He is one of the pioneers of neural networks, and was part of the small group of academics that nursed the technology through a period of tepid interest, funding, and development.

Amazon has been using algorithms to try to sell you extra stuff for years. But the technology to personalise merchandising, much further than recommendations, is advancing rapidly across ecommerce. Companies such as Sentient and Apptus and their AI-powered systems are changing site search functionality, product lists, facets and more, to try to generate more sales. I caught up with Sören Meelby, VP Marketing at Apptus, to get an introduction to the technology (Apptus eSales), and to pose some questions about the user experience in online retail. Sören Meelby: Each and every sort order typically follows a logic or business rules and'most popular' is fairly straight forward.

AMLG: Gary I'm super excited to have you today, thanks for coming on the show. We first met a few years ago in New York when I was running a tech meetup, the Singularity society, and you kindly came and spoke. You've been a professor of psychology at NYU for many years where your work has focused on language, biology, and the human mind. You've spent decades studying how children learn, and then in 2015 you founded this startup called Geometric Intelligence, focused on mining cognitive psychology for insights into building better machine learning techniques. Just this past December you were acquired by Uber to run their newly founded AI labs -- congratulations on that exit. So your algorithms offer an alternative approach to what is now a very popular branch of machine learning, called deep learning. Let's talk about deep learning -- it's a sexy buzzword which is thrown into about every startup pitch I see these days, and many corporate presentations, so I'm sure listeners have heard the term. What it really is is a rebranding of an old technique of using neural nets, which dates back to the 50s. Neural nets basically mimic the human neocortex, and by feeding in massive amounts, gigabytes of data and using tons of computational power, the algorithms are able to recognize patterns. Part of the reason why this technique is back in vogue is the combination of increasingly powerful computers combined with the massive training datasets that companies are building up. So there's been a flurry of activity, and the Googles and Facebooks of the world are throwing resources at the technique. As just one example, Facebook, using the over 400 billion photos people have uploaded, has built something called DeepFace, an image recognition tool that's now better than humans at recognizing whether two different images are of the same person. Gary you are well known as a critic of this technique, you've said that it's over-hyped. That there's some low hanging fruit that deep learning's good at -- specific narrow tasks like perception and categorization, and maybe beating humans at chess, but you felt that this deep learning mania was taking the field of AI in the wrong direction, that we're not making progress on cognition and strong AI. Or as you've put it, "we wanted Rosie the robot, and instead we got the roomba."

Providing a broad but in-depth introduction to neural network and machine learning in a statistical framework, this book provides a single, comprehensive resource for study and further research. All the major popular neural network models and statistical learning approaches are covered with examples and exercises in every chapter to develop a practical working understanding of the content. Each of the twenty-five chapters includes state-of-the-art descriptions and important research results on the respective topics. The broad coverage includes the multilayer perceptron, the Hopfield network, associative memory models, clustering models and algorithms, the radial basis function network, recurrent neural networks, principal component analysis, nonnegative matrix factorization, independent component analysis, discriminant analysis, support vector machines, kernel methods, reinforcement learning, probabilistic and Bayesian networks, data fusion and ensemble learning, fuzzy sets and logic, neurofuzzy models, hardware implementations, and some machine learning topics. Applications to biometric/bioinformatics and data mining are also included.