The artificial intelligence boom has expanded into creative fields once deemed uniquely human, like music, poetry, and even narrative podcasts. AI has also started writing rudimentary news articles and assisting reporters, but a new startup launched Wednesday says it will use AI to publish breaking news about a wide variety of topics. The site is called "Knowhere," and its creators say that they believe AI can be used to write unbiased news. The site will publish three versions of every article, aggregated from right-, left-, and center-leaning websites. "Fake news, the Russia misinformation scandal, and all of these issues that are at the top of the Zeitgeist at the moment are all symptoms of a fundamental problem that information moves too fast and at too large a scale for us to be able to reliably parse it and understand the world as human beings," Knowhere editor-in-chief and cofounder Nathaniel Barling and told me on the phone.

Porsche has bought a minority stake in Tel Aviv-based startup Anagog to expand its expertise in digital technology. Anagog has developed software that analyses user behavior directly in the mobile phone, using sensors, and then predicts future scenarios on the basis of artificial intelligence. In February, Daimler took part in a round of financing for Anagog. Last year, Anagog and Daimler launched the EQ Ready App, which helps drivers decide whether it makes sense for them to switch to greener car technology by recording real journeys and comparing them with electric and hybrid vehicles. "We are keen to work with new companies to continue developing new digital offerings and launch these solutions in the market quickly so that customers can start benefiting from them," Thilo Koslowski, Managing Director of Porsche Digital, a unit of Volkswagen Group's Porsche sports-car business, said in a statement on Wednesday.

In 2017, there were 7,672 games released on Steam, the world's most popular video game service. As of January 2018, if you filter out the dross, the average Steam game sells about 1,000 copies, according to independent data-crunching by publisher No More Robots. This is the dispiriting reality that independent game developers are working with: at this month's Game Developers Conference in San Francisco, there were queues up and down the hallway for a talk entitled "Making indie games that sell". Patrick Söderlund is the executive vice-president of one of the biggest game publishers in the world, EA – a company far removed from these problems, making multimillion-dollar games with studios that employ hundreds of people. Twenty years ago, though, he was part of a three-person team working out of an apartment in Sweden, making the original Battlefield games that have since become one of EA's largest money-makers.

Artificial intelligence researchers from nearly 30 countries are boycotting a South Korean university over concerns a new lab in partnership with a leading defence company could lead to "killer robots". More than 50 leading academics signed the letter calling for a boycott of Korea Advanced Institute of Science and Technology (KAIST) and its partner, defence manufacturer Hanwha Systems. The researchers said they would not collaborate with the university or host visitors from KAIST over fears it sought to "accelerate the arms race to develop" autonomous weapons. "There are plenty of great things you can do with AI that save lives, including in a military context, but to openly declare the goal is to develop autonomous weapons and have a partner like this sparks huge concern," said Toby Walsh, the organiser of the boycott and a professor at the University of New South Wales. "This is a very respected university partnering with a very ethically dubious partner that continues to violate international norms."

Motivation: Biomedical named entity recognition (BioNER) is the most fundamental task in biomedical text mining. State-of-the-art BioNER systems often require handcrafted features specifically designed for each type of biomedical entities. This feature generation process requires intensive labors from biomedical and linguistic experts, and makes it difficult to adapt these systems to new biomedical entity types. Although recent studies explored using neural network models for BioNER to free experts from manual feature generation, these models still require substantial human efforts to annotate massive training data. Results: We propose a multi-task learning framework for BioNER that is based on neural network models to save human efforts. We build a global model by collectively training multiple models that share parameters, each model capturing the characteristics of a different biomedical entity type. In experiments on five BioNER benchmark datasets covering four major biomedical entity types, our model outperforms state-of-the-art systems and other neural network models by a large margin, even when only limited training data are available. Further analysis shows that the large performance gains come from sharing character- and word-level information between different biomedical entities. The approach creates new opportunities for text-mining approaches to help biomedical scientists better exploit knowledge in biomedical literature.

Learning latent variable models with stochastic variational inference is challenging when the approximate posterior is far from the true posterior, due to high variance in the gradient estimates. We propose a novel rejection sampling step that discards samples from the variational posterior which are assigned low likelihoods by the model. Our approach provides an arbitrarily accurate approximation of the true posterior at the expense of extra computation. Using a new gradient estimator for the resulting unnormalized proposal distribution, we achieve average improvements of 3.71 nats and 0.21 nats over state-of-the-art single-sample and multi-sample alternatives respectively for estimating marginal log-likelihoods using sigmoid belief networks on the MNIST dataset.

The ability to identify sentiment in text, referred to as sentiment analysis, is one which is natural to adult humans. This task is, however, not one which a computer can perform by default. Identifying sentiments in an automated, algorithmic manner will be a useful capability for business and research in their search to understand what consumers think about their products or services and to understand human sociology. Here we propose two new Genetic Algorithms (GAs) for the task of automated text sentiment analysis. The GAs learn whether words occurring in a text corpus are either sentiment or amplifier words, and their corresponding magnitude. Sentiment words, such as 'horrible', add linearly to the final sentiment. Amplifier words in contrast, which are typically adjectives/adverbs like 'very', multiply the sentiment of the following word. This increases, decreases or negates the sentiment of the following word. The sentiment of the full text is then the sum of these terms. This approach grows both a sentiment and amplifier dictionary which can be reused for other purposes and fed into other machine learning algorithms. We report the results of multiple experiments conducted on large Amazon data sets. The results reveal that our proposed approach was able to outperform several public and/or commercial sentiment analysis algorithms.

Predictive modelling is a very exciting field with many different applications. Lots of algorithms have been developed in this area. According to many Kaggle competitions (Fogg, 2016), winning solutions are often obtained with elastic tools like random forest, gradient boosting or neural networks. These algorithms have many strengths but also share a major weakness, which is the lack of interpretability of a model structure. A single random forest, an xgboost model or a neural network may be parametrized with thousands of parameters which makes these models hard to understand.

We present a method for scalable and fully 3D magnetic field simultaneous localisation and mapping (SLAM) using local anomalies in the magnetic field as a source of position information. These anomalies are due to the presence of ferromagnetic material in the structure of buildings and in objects such as furniture. We represent the magnetic field map using a Gaussian process model and take well-known physical properties of the magnetic field into account. We build local magnetic field maps using three-dimensional hexagonal block tiling. To make our approach computationally tractable we use reduced-rank Gaussian process regression in combination with a Rao--Blackwellised particle filter. We show that it is possible to obtain accurate position and orientation estimates using measurements from a smartphone, and that our approach provides a scalable magnetic SLAM algorithm in terms of both computational complexity and map storage.

We conduct a large-scale study of language models for chord prediction. Specifically, we compare N-gram models to various flavours of recurrent neural networks on a comprehensive dataset comprising all publicly available datasets of annotated chords known to us. This large amount of data allows us to systematically explore hyper-parameter settings for the recurrent neural networks---a crucial step in achieving good results with this model class. Our results show not only a quantitative difference between the models, but also a qualitative one: in contrast to static N-gram models, certain RNN configurations adapt to the songs at test time. This finding constitutes a further step towards the development of chord recognition systems that are more aware of local musical context than what was previously possible.