Industrial Analytics (IA) describes the collection, analysis and usage of data generated in industrial operations and throughout the entire product lifecycle, applicable to any company that is manufacturing and selling physical products. It involves traditional methods of data capture and statistical modeling. However, most of its future value will be enabled by advancements in connectivity (IoT) and improved methods for analyzing and interpreting data (Machine Learning). A fascinating report on how Industrial Analytics is maturing based on advances in the areas of IoT, machine learning, and big data analytics was published this week. This study was initiated and governed by the Digital Analytics Association e.V. Germany (DAAG), which runs a professional working group on the topic of Industrial Analytics.

We're hearing about a lot of companies using artificial intelligence (AI) to make the most of the data they collect. Now market research has adopted the technology. After finding success with clients such as Google and Mastercard abroad, Prague-based response:now is bringing their AI-powered app to the United States. The company now offers a fully self-service, programmatic platform that creates research reports based on machine learning. Then it uses a human editor to tease out any undetected nuances and reconcile any disparities.

Any time of year is the perfect time to buy a robot for yourself or someone who needs more robots in their life, but this particular time of year is even perfecter than most: The holidays are approaching, all kinds of things are on sale, and nobody will ask questions if a whole bunch of new robots suddenly show up in your house. To help you decide which robots to buy for yourself and which to buy for yourself and for other people, we've put together a brand new edition of our annual Robots Gift Guide. It's stuffed with giftable robots ranging from affordable to ridiculous, and we promise that if you don't find something you like, we'll feel bad about it and be sad. Also, don't forget that we've got robot gift guides going back like five years (here: 2016, 2015, 2014, 2013, 2012), and since we try to mix them up every year, they're great places for even more ideas for robots that are probably way cheaper now than when we first posted about them. And remember: While we provide prices and links to places where you can buy these items, we're not endorsing any in particular, and a little bit of searching may result in better deals (all prices are in U.S. dollars).

Bayesian optimization is a sample-efficient approach to solving global optimization problems. Along with a surrogate model, this approach relies on theoretically motivated value heuristics (acquisition functions) to guide the search process. Maximizing acquisition functions yields the best performance; unfortunately, this ideal is difficult to achieve since optimizing acquisition functions per se is frequently non-trivial. This statement is especially true in the parallel setting, where acquisition functions are routinely non-convex, high-dimensional, and intractable. Here, we demonstrate how many popular acquisition functions can be formulated as Gaussian integrals amenable to the reparameterization trick and, ensuingly, gradient-based optimization. Further, we use this reparameterized representation to derive an efficient Monte Carlo estimator for the upper confidence bound acquisition function in the context of parallel selection.

In this paper, we study the effects of different prior and likelihood choices for Bayesian matrix factorisation, focusing on small datasets. These choices can greatly influence the predictive performance of the methods. We identify four groups of approaches: Gaussian-likelihood with real-valued priors, nonnegative priors, semi-nonnegative models, and finally Poisson-likelihood approaches. For each group we review several models from the literature, considering sixteen in total, and discuss the relations between different priors and matrix norms. We extensively compare these methods on eight real-world datasets across three application areas, giving both inter- and intra-group comparisons. We measure convergence runtime speed, cross-validation performance, sparse and noisy prediction performance, and model selection robustness. We offer several insights into the trade-offs between prior and likelihood choices for Bayesian matrix factorisation on small datasets - such as that Poisson models give poor predictions, and that nonnegative models are more constrained than real-valued ones.

End-to-end neural network based approaches to audio modelling are generally outperformed by models trained on high-level data representations. In this paper we present preliminary work that shows the feasibility of training the first layers of a deep convolutional neural network (CNN) model to learn the commonly-used log-scaled mel-spectrogram transformation. Secondly, we demonstrate that upon initializing the first layers of an end-to-end CNN classifier with the learned transformation, convergence and performance on the ESC-50 environmental sound classification dataset are similar to a CNN-based model trained on the highly pre-processed log-scaled mel-spectrogram features.

In this paper, we study the optimal convergence rate for distributed convex optimization problems in networks. We model the communication restrictions imposed by the network as a set of affine constraints and provide optimal complexity bounds for four different setups, namely: the function $F(\xb) \triangleq \sum_{i=1}^{m}f_i(\xb)$ is strongly convex and smooth, either strongly convex or smooth or just convex. Our results show that Nesterov's accelerated gradient descent on the dual problem can be executed in a distributed manner and obtains the same optimal rates as in the centralized version of the problem (up to constant or logarithmic factors) with an additional cost related to the spectral gap of the interaction matrix. Finally, we discuss some extensions to the proposed setup such as proximal friendly functions, time-varying graphs, improvement of the condition numbers.

Computers have become much more adept at translating from one language into another in recent years, thanks to the application of neural networks. However, these AI systems usually require a lot of content translated by humans for the computers to learn from, while two new papers demonstrate that it's possible to develop a system that doesn't rely on parallel texts. Mikel Artetxe, a computer scientist at the University of the Basque Country (UPV) and the author of one of these papers, compares the situation to giving someone various books in Chinese and various books in Arabic, without any of the same texts overlapping. A human would find it very difficult to learn how to translate from Chinese into Arabic in this scenario, but a computer might not. In a typical machine-learning process, the AI system would be supervised.

A United Nations meeting in Geneva earlier this month on lethal autonomous weapons systems (LAWS) was derailed when Russia said they would not adhere to any prohibitions on killer robots, according to Defense One. The U.N. meeting appeared to be undermined both by Russia's disinterest in it and the framework of the meeting itself. Member nations attempted to come in and define what LAWS' systems would be, and what restrictions could be developed around autonomous war machines, but no progress was made. In a statement, Russia said that the lack of already developed war machines makes coming up with prohibitions on such machines difficult. "According to the Russian Federation, the lack of working samples of such weapons systems remains the main problem in the discussion on LAWS…this can hardly be considered as an argument for taking preventive prohibitive or restrictive measures against LAWS being a by far more complex and wide class of weapons of which the current understanding of humankind is rather approximate," read the statement.

Pioneering Blockchain Artificial Intelligence (AI) marketplace SingularityNET has announced a $36 mln ICO to fund its expansion. Announcing the fairly modest funding goal for the world's first such AI marketplace, SingularityNET's robotic'chief humanoid officer' confirmed the 500 mln token event at its recent Web Summit in Lisbon, Portugal. A combination of private and public sales will distribute the tokens, with the ICO concluding when all are sold or when the $36 mln figure has been reached. Participants are required to register on a whitelist prior to the sale beginning Dec. 8. "The exact amount of tokens available in the crowdsale depends on private sales, which have not yet been finalized," an accompanying press release states.