This paper extends the recently proposed and theoretically justified iterative thresholding and $K$ residual means algorithm ITKrM to learning dicionaries from incomplete/masked training data (ITKrMM). It further adapts the algorithm to the presence of a low rank component in the data and provides a strategy for recovering this low rank component again from incomplete data. Several synthetic experiments show the advantages of incorporating information about the corruption into the algorithm. Finally, image inpainting is considered as application example, which demonstrates the superior performance of ITKrMM in terms of speed at similar or better reconstruction quality compared to its closest dictionary learning counterpart.

We propose a generic framework based on a new stochastic variance-reduced gradient descent algorithm for accelerating nonconvex low-rank matrix recovery. Starting from an appropriate initial estimator, our proposed algorithm performs projected gradient descent based on a novel semi-stochastic gradient specifically designed for low-rank matrix recovery. Based upon the mild restricted strong convexity and smoothness conditions, we derive a projected notion of the restricted Lipschitz continuous gradient property, and prove that our algorithm enjoys linear convergence rate to the unknown low-rank matrix with an improved computational complexity. Moreover, our algorithm can be employed to both noiseless and noisy observations, where the optimal sample complexity and the minimax optimal statistical rate can be attained respectively. We further illustrate the superiority of our generic framework through several specific examples, both theoretically and experimentally.

We propose a novel high-dimensional linear regression estimator: the Discrete Dantzig Selector, which minimizes the number of nonzero regression coefficients subject to a budget on the maximal absolute correlation between the features and residuals. Motivated by the significant advances in integer optimization over the past 10-15 years, we present a Mixed Integer Linear Optimization (MILO) approach to obtain certifiably optimal global solutions to this nonconvex optimization problem. The current state of algorithmics in integer optimization makes our proposal substantially more computationally attractive than the least squares subset selection framework based on integer quadratic optimization, recently proposed in [8] and the continuous nonconvex quadratic optimization framework of [33]. We propose new discrete first-order methods, which when paired with state-of-the-art MILO solvers, lead to good solutions for the Discrete Dantzig Selector problem for a given computational budget. We illustrate that our integrated approach provides globally optimal solutions in significantly shorter computation times, when compared to off-the-shelf MILO solvers. We demonstrate both theoretically and empirically that in a wide range of regimes the statistical properties of the Discrete Dantzig Selector are superior to those of popular $\ell_{1}$-based approaches. We illustrate that our approach can handle problem instances with p = 10,000 features with certifiable optimality making it a highly scalable combinatorial variable selection approach in sparse linear modeling.

We propose a new family of message passing techniques for MAP estimation in graphical models which we call {\em Sequential Reweighted Message Passing} (SRMP). Special cases include well-known techniques such as {\em Min-Sum Diffusion} (MSD) and a faster {\em Sequential Tree-Reweighted Message Passing} (TRW-S). Importantly, our derivation is simpler than the original derivation of TRW-S, and does not involve a decomposition into trees. This allows easy generalizations. We present such a generalization for the case of higher-order graphical models, and test it on several real-world problems with promising results.

January 18, 2017 --Barren circles dot the dry grasslands across about 1,500 miles of the Namib Desert stretching down the southwestern coast of Africa, emerging, growing, shrinking, and disappearing in lifetimes of 30 to 60 years. The empty patches are accentuated by a rim of particularly tall grasses that ring the circles, which range from 6 feet to 115 feet wide. The fairy circles, as the strange bare soil spots are called, have long puzzled scientists. Although they look a bit like imprints left by massive raindrops, impacting meteors, or as legend would have it, the feet of gods, researchers suspect the pattern may form as a result of a more systematic natural process. But just what that process might be has been the subject of much debate.

According to a recent press release out of the company, "Deloitte predicts that over 300 million smartphones, or more than one-fifth of units sold in 2017, will have machine learning capabilities within the device in the next 12 months. The 16th edition of the'Technology, Media & Telecommunications (TMT) Predictions' showcases how mobile devices will be able to perform machine learning tasks even without connectivity, which will significantly alter how humans interact with technology across every industry, market and society. However, over time machine learning on-the-go will not just be limited to smartphones. These capabilities are likely to be found in tens of millions (or more) of drones, tablets, cars, virtual or augmented reality devices, medical tools, Internet of Things (IoT) devices and unforeseen new technologies."

Add artificial intelligence (AI) into the equation--and more than a few apocalyptic movies about such learning-enabled machines taking over the human race--and the fear factor is ratcheted up a bit. Though ethical concerns are prevalent, executives across the globe say AI is nonetheless certain to develop further in their businesses. According to a study released this week by Infosys, 71 percent of the 1,600 senior business decision-makers surveyed say that the rise of AI in the workplace is inevitable, pointing to positive changes for business prospects, employees and society. The study--commissioned by Infosys and conducted by independent market researcher Vanson Bourne--set out to investigate the approach and attitudes that senior decision-makers in large organizations (at least 1,000 employees and $500 million in annual revenue) have toward AI technology and how they see the future application and development of AI in their industries. Although AI definitions can vary, it is generally considered as an activity traditionally performed through human intelligence that can now be done by a computer.

The iPhone 8 could include face recognition and a "wraparound" screen design, analyst Timothy Arcuri of Cowen and Company said, according to Business Insider. Arcuri predicts three iPhone models later this year, according to a research note circulated to Cowen and Company clients. The next iPhone 8 is referred to in the note as the "iPhone X." The note says one of the models, the iPhone X, will be a 5.8 inch OLED iPhone 8 with a "wraparound" "fixed flex" screen design with embedded sensors, according to Apple Insider, who also obtained the note. The model is rumored to come with features such as, face recognition.

As the artificial intelligence space in Montreal heats up, a group of McGill student entrepreneurs are set to launch the McGill Artificial Intelligence Society. The McGill AI Society describes themselves as "a community of passionate students [who are] all about learning, practicing and sharing knowledge of the increasingly interesting field of AI." Led by Théo Szymkowiak, a third year McGill Computer Science student and former CTO of Fractal (McGill X-1 Cohort 2016), the McGill AI Society aims to provide introductory and advanced classes on Machine Learning and Artificial Intelligence where members will learn how to build robots, code up image classifiers, learn about speech recognition, and much more. McGill AI Society members will also have the opportunity to present their work and teach other students as well. This society is open to all McGill students from every faculty. Interested McGill students can register to be part of the McGill AI Society here.

Do you think AI will decrease human labor? So if foreseeable technologies materialize, then then the need for human labor could decrease. Technology always puts existing jobs under strain. This doesn't immediately mean that human labor as a whole is under threat. Generally, other professions grow to fill the loss, often creating more jobs than the ones that are lost.