Many computationally-efficient methods for Bayesian deep learning rely on continuous optimization algorithms, but the implementation of these methods requires significant changes to existing code-bases. In this paper, we propose Vprop, a method for Gaussian variational inference that can be implemented with two minor changes to the off-the-shelf RMSprop optimizer. Vprop also reduces the memory requirements of Black-Box Variational Inference by half. We derive Vprop using the conjugate-computation variational inference method, and establish its connections to Newton's method, natural-gradient methods, and extended Kalman filters. Overall, this paper presents Vprop as a principled, computationally-efficient, and easy-to-implement method for Bayesian deep learning.

Machine learning approaches hold great potential for the automated detection of lung nodules in chest radiographs, but training the algorithms requires vary large amounts of manually annotated images, which are difficult to obtain. Weak labels indicating whether a radiograph is likely to contain pulmonary nodules are typically easier to obtain at scale by parsing historical free-text radiological reports associated to the radiographs. Using a repositotory of over 700,000 chest radiographs, in this study we demonstrate that promising nodule detection performance can be achieved using weak labels through convolutional neural networks for radiograph classification. We propose two network architectures for the classification of images likely to contain pulmonary nodules using both weak labels and manually-delineated bounding boxes, when these are available. Annotated nodules are used at training time to deliver a visual attention mechanism informing the model about its localisation performance. The first architecture extracts saliency maps from high-level convolutional layers and compares the estimated position of a nodule against the ground truth, when this is available. A corresponding localisation error is then back-propagated along with the softmax classification error. The second approach consists of a recurrent attention model that learns to observe a short sequence of smaller image portions through reinforcement learning. When a nodule annotation is available at training time, the reward function is modified accordingly so that exploring portions of the radiographs away from a nodule incurs a larger penalty. Our empirical results demonstrate the potential advantages of these architectures in comparison to competing methodologies.

Learning attribute applicability of products in the Amazon catalog (e.g., predicting that a shoe should have a value for size, but not for battery-type) at scale is a challenge. The need for an interpretable model is contingent on (1) the lack of ground truth training data, (2) the need to utilise prior information about the underlying latent space and (3) the ability to understand the quality of predictions on new, unseen data. To this end, we develop the MaxMachine, a probabilistic latent variable model that learns distributed binary representations, associated to sets of features that are likely to cooccur in the data. Layers of MaxMachines can be stacked such that higher layers encode more abstract information. Any set of variables can be clamped to encode prior information. We develop fast sampling based posterior inference. Preliminary results show that the model improves over the baseline in 17 out of 19 product groups and provides qualitatively reasonable predictions.

The problem of universal outlying sequence detection is studied, where the goal is to detect outlying sequences among $M$ sequences of samples. A sequence is considered as outlying if the observations therein are generated by a distribution different from those generating the observations in the majority of the sequences. In the universal setting, we are interested in identifying all the outlying sequences without knowing the underlying generating distributions. In this paper, a class of tests based on distribution clustering is proposed. These tests are shown to be exponentially consistent with linear time complexity in $M$. Numerical results demonstrate that our clustering-based tests achieve similar performance to existing tests, while being considerably more computationally efficient.

A popular argument for the simulation hypothesis came from University of Oxford philosopher Nick Bostrom in 2003, when he suggested that members of an advanced civilization with enormous computing power might decide to run simulations of their ancestors. They would probably have the ability to run many, many such simulations, to the point where the vast majority of minds would actually be artificial ones within such simulations, rather than the original ancestral minds. So simple statistics suggest it is much more likely that we are among the simulated minds. And there are other reasons to think we might be virtual. For instance, the more we learn about the universe, the more it appears to be based on mathematical laws. Perhaps that is not a given, but a function of the nature of the universe we are living in.

'Fear and Loathing In the Age of QE … AI' is a presentation given at Mining Investment London earlier this week. Stephen Flood, CEO of GoldCore presentation (28 minutes) was well received at the conference which is a strategic mining and investment conference for leaders in the mining and investment sectors, bringing together attendees from 20 countries. 'Fear and Loathing In the Age of QE … AI' can be watched on Youtube here Why Silver Bullion Is Set To Soar – GoldCore Interview Gold Bullion Stored In Singapore Is Safest – Marc Faber Russia Seen More Likely to Sell Dollar Rather Than Gold Talking Gold with CNN's Richard Quest Gold holds near one-week low as dollar firms (Reuters.com) Goldman Says the Bitcoin Haters Just Don't Get It (Bloomberg.com) Goldman Warns That Market Valuations Are at Their Highest Since 1900 (Bloomberg.com)

In recent years, the media has devoted a lot of time and space to how robots and machines are taking on more and more human jobs. The stories often give rise to fear and a negative sense of what artificial intelligence and machine learning could potentially do. However, away from the more sensational headlines, good news emerges of how machines can help humans make businesses more efficient, transparent and cost effective. It's no surprise to hear that SMEs are the engine of any country's economy. In addition, the World Bank states that the 600 million jobs which need to be created to absorb a growing global workforce will have to come from SMEs.

In digital marketing and SEO there is no one-size-fits-all strategy or a formula for a guaranteed success. To make the best decisions for your company, it's invaluable to follow the trends and listen to established voices of the industry. Further we offer a short interview with one of the SEO influencers that has some useful insights to share. Maria knows 18 languages which has made her a multilingual SEO, PPC and social media marketing expert. She has managed software projects for well-known IT companies and banks, as well as cooperated with governments and police authorities.

Facebook has revealed more about how it is now using artificial intelligence to remove terror related content that appears on its platforms. The tech giant has come under pressure from the UK government on the matter, with Prime Minister Theresa May going as far as accusing Facebook and others of providing a "safe space" for terrorists. The social network said in an update on its efforts today that it was "hopeful AI will become a more important tool in the arsenal of protection and safety on the internet and on Facebook" and demonstrated the success of its early efforts. It said that 99 per cent of content related to ISIS and Al Qaeda that's removed is identified using AI before it is flagged by humans. However, it noted that AI could not be a silver bullet and human efforts were still required.

Researcher groups at the University of the Basque Country in Spain, and at Facebook, have separately developed unsupervised machine-learning techniques for teaching neural networks to translate between languages without requiring parallel texts. Researchers at the University of the Basque Country (UPV) in Spain and Facebook have separately developed unsupervised machine-learning techniques for teaching neural networks to translate between languages with no parallel texts. Each method employs as training strategies back translation and denoising; in the first process, a sentence in one language is approximately translated into the other, then translated back into the original language, with networks adjusted to make subsequent attempts closer to identical. Meanwhile, denoising adds noise to a sentence by rearranging or removing words, and attempts to translate that back into the original. The UPV method translates more frequently during training, while the Facebook technique, in addition to encoding a sentence from one language into a more abstract representation before decoding it into the other language, also confirms the intermediate language is truly abstract.