The World Cup 2018 has officially begun--which means, if you're a hardcore soccer fan, you're pretty tied up for the next month watching the matches. For those who can't or don't want to follow the action, here's a major spoiler: Germany is going to beat Vegas-favorite Brazil in the final, and Spain and France will round out the tournament's final four teams. That prediction comes courtesy of Unanimous A.I., an artificial intelligence startup that performs a kind of complex crowdsourcing. Founded by scientist and engineer Louis Rosenberg, Unanimous can be used to better understand the nuanced opinions of a population, which makes it useful for tasks like performing market research, diagnosing diseases, or making predictions about the future. Launched in 2014, the company's technology already has an impressive résumé of accurate forecasts.

AI is becoming so common that it's almost boring. While we're seeing some surprising new projects using AI, including detecting movement through walls and predicting that Brazil will be the winner of this year's World Cup, other companies are using the tech for more pedestrian applications, like improving translation and photo retouching. Now, Microsoft has an AI-powered system that decides which Windows 10 devices should get an Update first. The company says that it is "seeing higher satisfaction numbers, fewer known issues, and lower support call volumes compared to previous Windows 10 releases." Microsoft piloted the program during last Fall's Creator's Update rollout, gathering data on the types of Windows 10 devices that had good update experiences.

Gibbs sampling is a Markov chain Monte Carlo method that is one of the most widespread techniques used with graphical models [7]. Gibbs sampling is an iterative method that repeatedly resamples a variable in the model from its conditional distribution, a process that is guaranteed to converge asymptotically to the desired distribution. Since these updates are typically simple and fast to run, Gibbs sampling can be applied to a variety of problems, and has been used for inference on large-scale graphical models in many systems [11, 13, 14, 19, 20, 21]. Unfortunately, for large graphical models with many factors, the computational cost of running an iteration of Gibbs sampling can become prohibitive. Even though Gibbs sampling is a graph-local algorithm, in the sense that each update only needs to reference data associated with a local neighborhood of the factor graph, as graphs become large and highly connected, even these local neighborhoods can become huge.

We consider the problem of learning a low-rank matrix, constrained to lie in a linear subspace, and introduce a novel factorization for modeling such matrices. A salient feature of the proposed factorization scheme is it decouples the low-rank and the structural constraints onto separate factors. We formulate the optimization problem on the Riemannian spectrahedron manifold, where the Riemannian framework allows to develop computationally efficient conjugate gradient and trust-region algorithms. Experiments on problems such as standard/robust/nonnegative matrix completion, Hankel matrix learning and multi-task learning demonstrate the efficacy of our approach. A shorter version of this work has been published in ICML'18 (Jawanpuria and Mishra, 2018).

In 2010, an octopus named Paul picked the winning team in 12 out of 14 matches World Cup matches, including the final. But it's not 2010 anymore; we no longer need an octopus, regardless of how good he is at his job. In 2018, a group of researchers from Germany and Belgium felt it best to let computers do the heavy lifting. The group built a model using a number of factors, such as FIFA rankings, population, gross domestic product (GDP), the number of players who play together on a single club, average age of a club's players, and how many Champions League finals each has won. The team then paired that data with betting odds from the larger bookmakers and ran the simulation 100,000 times to try and pick a winner.

An artificial intelligence machine ran 100,000 simulations of the 2018 FIFA World Cup, which is hosted in Russia, and concluded that the winner of the international sporting event will be Spain. If Spain is not to prevail, the machine sequentially picked Germany, Brazil, France, Belgium, or Argentina to win. The calculations, which were developed by a group of researchers in Germany and Belgium, used a number of factors to determine the winner. They included, but weren't limited to, FIFA rankings, population, gross domestic product (GDP), the number of players who play together on a single club, average age of a club's players, and how many Champions League finals each has won. The team, according to The Next Web, proceeded to pair that data with betting odds from the larger bookmakers and ran the simulation 100,000 times to try and pick the victorious team(s). However, the researchers told Motherboard that given "the myriad of possible constellations this exact tournament course is still extremely unlikely."

South African-based data scientists at Principa are at it again; this time using predictive analytics and machine learning to foretell the results of the 2018 Football World Cup. The 2018 FIFA World Cup kicks off tomorrow in Russia with the host nation taking on Saudi Arabia in Group A. Principa has already predicted the results for all the first games in the first round of matches. The company's data scientists use different algorithms to develop models that can predict the outcome of the matches. Principa notes that as the objective of machine learning is to develop models that can retrain themselves to adapt when exposed to new data, the algorithms will be re-trained with the results of each match to improve the accuracy of the following round's generated prediction. It points out that the purpose is to see how well different predictive analytics techniques used successfully in other areas can outperform the best human-made predictions.

A platform that can diagnose several diseases with a high degree of precision using metabolic markers found in patients' blood has been developed by scientists at the University of Campinas (UNICAMP) in Brazil. The method combines mass spectrometry, which can identify tens of thousands of molecules present in blood serum, with an artificial intelligence algorithm capable of finding patterns associated with diseases of viral, bacterial, fungal and even genetic origin. The results have been published in Frontiers in Bioengineering and Biotechnology. "We used infection by Zika virus as a model to develop the platform and showed that in this case, diagnostic accuracy exceeded 95%. One of the main advantages is that the method doesn't lose sensitivity even if the virus mutates," said Melo's supervisor Rodrigo Ramos Catharino, principal investigator for the project.

In the past few years, Convolutional Neural Networks (CNNs) have been achieving state-of-the-art performance on a variety of problems. Many companies employ resources and money to generate these models and provide them as an API, therefore it is in their best interest to protect them, i.e., to avoid that someone else copies them. Recent studies revealed that state-of-the-art CNNs are vulnerable to adversarial examples attacks, and this weakness indicates that CNNs do not need to operate in the problem domain (PD). Therefore, we hypothesize that they also do not need to be trained with examples of the PD in order to operate in it. Given these facts, in this paper, we investigate if a target black-box CNN can be copied by persuading it to confess its knowledge through random non-labeled data. The copy is two-fold: i) the target network is queried with random data and its predictions are used to create a fake dataset with the knowledge of the network; and ii) a copycat network is trained with the fake dataset and should be able to achieve similar performance as the target network. This hypothesis was evaluated locally in three problems (facial expression, object, and crosswalk classification) and against a cloud-based API. In the copy attacks, images from both non-problem domain and PD were used. All copycat networks achieved at least 93.7% of the performance of the original models with non-problem domain data, and at least 98.6% using additional data from the PD. Additionally, the copycat CNN successfully copied at least 97.3% of the performance of the Microsoft Azure Emotion API. Our results show that it is possible to create a copycat CNN by simply querying a target network as black-box with random non-labeled data.

This paper develops a computationally efficient posterior sampling algorithm for Bayesian linear regression models with Gaussian errors. Our new approach is motivated by the fact that existing software implementations for Bayesian linear regression do not readily handle problems with large number of observations (hundreds of thousands) and predictors (thousands). Moreover, existing sampling algorithms for popular shrinkage priors are bespoke Gibbs samplers based on case-specific latent variable representations. By contrast, the new algorithm does not rely on case-specific auxiliary variable representations, which allows for rapid prototyping of novel shrinkage priors outside the conditionally Gaussian framework. Specifically, we propose a slice-within-Gibbs sampler based on the elliptical slice sampler of Murray et al. [2010].