Acquisition cost is a crucial bottleneck for seismic workflows, and low-rank formulations for data interpolation allow practitioners to `fill in' data volumes from critically subsampled data acquired in the field. Tremendous size of seismic data volumes required for seismic processing remains a major challenge for these techniques. We propose a new approach to solve residual constrained formulations for interpolation. We represent the data volume using matrix factors, and build a block-coordinate algorithm with constrained convex subproblems that are solved with a primal-dual splitting scheme. The new approach is competitive with state of the art level-set algorithms that interchange the role of objectives with constraints. We use the new algorithm to successfully interpolate a large scale 5D seismic data volume, generated from the geologically complex synthetic 3D Compass velocity model, where 80% of the data has been removed.

We consider a contextual version of multi-armed bandit problem with global knapsack constraints. In each round, the outcome of pulling an arm is a scalar reward and a resource consumption vector, both dependent on the context, and the global knapsack constraints require the total consumption for each resource to be below some pre-fixed budget. The learning agent competes with an arbitrary set of context-dependent policies. This problem was introduced by Badanidiyuru et al. (2014), who gave a computationally inefficient algorithm with near-optimal regret bounds for it. We give a computationally efficient algorithm for this problem with slightly better regret bounds, by generalizing the approach of Agarwal et al. (2014) for the non-constrained version of the problem. The computational time of our algorithm scales logarithmically in the size of the policy space. This answers the main open question of Badanidiyuru et al. (2014). We also extend our results to a variant where there are no knapsack constraints but the objective is an arbitrary Lipschitz concave function of the sum of outcome vectors.

Quantile regression gives a comprehensive picture of the relationship between a response variable and a set of predictors. It is particularly appealing when the inferential interest lies in the probabilistic properties of extreme observations conditional on a set of predictors. Such objectives arise in various disciplines: in environmental sciences, Friederichs and Hense (2007) study the probabilistic properties of extreme precipitation events, while Pedersen (2015) model the tail distribution of stock and bond returns. In an epidemiological study, Burgette et al. (2011) use penalized quantile regression to explore covariates that affect the lower tail of the distribution of birth weight of babies. When the distribution of the dependent variable is skewed, the desire for robustness to extreme observations makes quantile regression a preferred approach. Examples include the study of tourist expense patterns in Marrocu et al. (2015) and wage distribution in Buchinsky (1995).

A significant portion of existing variable selection methods are only applicable to linear parametric models. Despite the linearity and additivity assumption, variable selection in linear regression models has been popular since 1970; refer to Akaike information criterion [AIC; Akaike (1973)]; Bayesian information criterion [BIC; Schwarz et al (1978)] and Risk inflation criterion [RIC; Foster and George (1994)]. Popular classical sparse-regression methods such as Least absolute shrinkage operator [LASSO; Tibshirani (1996); Efron et al (2004)], and related penalization methods (Fan and Li, 2001; Zou and Hastie, 2005; Zou, 2006; Zhang, 2010) have gained popularity over the last decade due to their simplicity, computational scalability and efficiency in prediction when the underlying relation between the response and the predictors can be adequately described by parametric models. Bayesian methods (Mitchell and Beauchamp, 1988; George and McCulloch, 1993, 1997) with sparsity inducing priors offers greater applicability beyond parametric models and are a convenient alternative when the underlying goal is in inference and uncertainty quantification. However, there is still a limited amount of literature which seriously considers relaxing the linearity assumption, particularly when the dimension of the predictors is high. Moreover, when the focus is on learning the interactions between the variables, parametric models are often restrictive since they require very many parameters to capture the higher-order interaction terms. 2 Smoothing based non-additive nonparametric regression methods (Lafferty and Wasser-man, 2008; Wahba, 1990; Green and Silverman, 1993; Hastie and Tibshirani, 1990) can accommodate a wide range of relationships between predictors and response leading to excellent predictive performance.

From Harry Potter and Romeo and Juliet to the stories of Oedipus and Icarus, almost every tale told conforms to one of just six plots, researchers have claimed. A major new analysis of over 1,700 stories identified the core plots'which form the building blocks of complex narratives'. Researchers used complex data-mining to locate words linked to positive or negative emotion in each story to reveal the set of arcs. A major new analysis of over 1,700 stories identified the core plots'which form the building blocks of complex narratives'. Shown, the plot of Harry Potter and the Deathly Hallows, which researchers found has the'rise, fall rise' plot.

Seattle-based machine learning startup Dato, which originally launched as GraphLab, announced today that it has changed its name again. Now, the company is known as Turi, an ode to computer science legend Alan Turing. This past September, Turi -- then Dato -- found itself in a trademark infringement argument with fellow tech startup Datto after Dato changed its name from GraphLab in January 2015. Datto, which has been offering data backup and recovery services from its Connecticut headquarters since 2007, first complained about Dato's new name just weeks after the company finished its rebranding from GraphLab. Datto argued that Dato's new name would cause confusion.

Scientists are closer to creating a computer with emotions. Researchers in Russia are expected to reveal an emotional computer within a year and a half, which will be able to think like a person and build up trust, its creators say. The system, called'Virtual Actor', is being created by the National Research Nuclear University in Moscow. Computers are machines used for practical reasons, without any emotion involved. The AI, called'Virtual Actor', is expected to be online within the next year and a half.

Being able to make near-real-time decisions becomes increasingly crucial. To succeed, we need machine learning systems that can turn massive amounts of data into valuable insights. But when you're just starting out in the data science field, how do you get started creating machine learning applications? The answer is TensorFlow, a new open source machine learning library from Google that they use in their own successful products like Search, Maps, YouTube, Translate, and Photos. The TensorFlow library can take your high level designs and turn them into the low level mathematical operations required by machine learning algorithms.

Over a couple of drinks, you cover the usual suspects: favourite foods, dream jobs, where you each grew up. On the way home, you wonder: Were they into me? They were smiling a lot, so probably. But they also looked at their watch a few times, so probably not. Now imagine that the whole time you two were talking, scientists were sitting under the table transcribing the conversation.

Hell, Microsoft issued an Education Edition of the popular PC game earlier this year targeted at use in a classroom setting. Turns out the game could also prove a useful tool for helping artificial intelligence be more, well, intelligent. Back in March, Microsoft Research showcased the work it was doing with Project Malmo, a platform designed to leverage Minecraft as a means of helping improve AI problem solving, using machines to accomplish tasks and create items in the blocky game. Now the company is bringing Malmo to the GitHub-using masses, courtesy of an open-source license in a private preview. Katja Hoffman of MS's Cambridge, UK lab highlighted the key of teaching AI fundamental connections that go build simple pattern recognition.