The goal of this paper is to improve learning for multivariate processes whose structure is dependent on some known graph topology; especially when the number of available samples is much smaller than the number of variables. Typically, the graph information is incorporated into the learning process via a smoothness assumption postulating that the values supported on well-connected vertices exhibit small variations. We argue that smoothness is not enough. To capture the behavior of complex interconnected systems, such as transportation and biological networks, it is important to train expressive models, being able to reproduce a wide range of graph and temporal behaviors. Motivated by this need, this paper puts forth a novel definition of time-vertex wide-sense stationarity, or joint stationarity for short. We believe that the proposed definition is natural, at it intimately relates to existing definitions of stationarity in the time and vertex domains. We use joint stationarity to regularize learning and to reduce computational complexity in both estimation and recovery tasks. In particular, we show that for any jointly stationary process: (a) one can learn the covariance structure from O(1) samples, and (b) can solve MMSE recovery problems, such as interpolation, denoising, forecasting, in complexity that is linear to the edges and timesteps. Experiments with three datasets suggest that joint stationarity can yield significant accuracy improvements in the reconstruction effort of under-sampled problems, even when the graph is only approximately known or the process is only close to stationary.

One of the longstanding open problems in spectral graph clustering (SGC) is the so-called model order selection problem: automated selection of the correct number of clusters. This is equivalent to the problem of finding the number of connected components or communities in an undirected graph. We propose automated model order selection (AMOS), a solution to the SGC model selection problem under a random interconnection model (RIM) using a novel selection criterion that is based on an asymptotic phase transition analysis. AMOS can more generally be applied to discovering hidden block diagonal structure in symmetric non-negative matrices. Numerical experiments on simulated graphs validate the phase transition analysis, and real-world network data is used to validate the performance of the proposed model selection procedure.

The generalization error of deep neural networks via their classification margin is studied in this work. Our approach is based on the Jacobian matrix of a deep neural network and can be applied to networks with arbitrary non-linearities and pooling layers, and to networks with different architectures such as feed forward networks and residual networks. Our analysis leads to the conclusion that a bounded spectral norm of the network's Jacobian matrix in the neighbourhood of the training samples is crucial for a deep neural network of arbitrary depth and width to generalize well. This is a significant improvement over the current bounds in the literature, which imply that the generalization error grows with either the width or the depth of the network. Moreover, it shows that the recently proposed batch normalization and weight normalization re-parametrizations enjoy good generalization properties, and leads to a novel network regularizer based on the network's Jacobian matrix. The analysis is supported with experimental results on the MNIST, CIFAR-10, LaRED and ImageNet datasets.

Injuries have a great impact on professional soccer, due to their large influence on team performance and the considerable costs of rehabilitation for players. Existing studies in the literature provide just a preliminary understanding of which factors mostly affect injury risk, while an evaluation of the potential of statistical models in forecasting injuries is still missing. In this paper, we propose a multidimensional approach to injury prediction in professional soccer which is based on GPS measurements and machine learning. By using GPS tracking technology, we collect data describing the training workload of players in a professional soccer club during a season. We show that our injury predictors are both accurate and interpretable by providing a set of case studies of interest to soccer practitioners. Our approach opens a novel perspective on injury prevention, providing a set of simple and practical rules for evaluating and interpreting the complex relations between injury risk and training performance in professional soccer.

In this paper we consider the cluster estimation problem under the Stochastic Block Model. We show that the semidefinite programming (SDP) formulation for this problem achieves an error rate that decays exponentially in the signal-to-noise ratio. The error bound implies weak recovery in the sparse graph regime with bounded expected degrees, as well as exact recovery in the dense regime. An immediate corollary of our results yields error bounds under the Censored Block Model. Moreover, these error bounds are robust, continuing to hold under heterogeneous edge probabilities and a form of the so-called monotone attack. Significantly, this error rate is achieved by the SDP solution itself without any further pre- or post-processing, and improves upon existing polynomially-decaying error bounds proved using the Grothendieck\textquoteright s inequality. Our analysis has two key ingredients: (i) showing that the graph has a well-behaved spectrum, even in the sparse regime, after discounting an exponentially small number of edges, and (ii) an order-statistics argument that governs the final error rate. Both arguments highlight the implicit regularization effect of the SDP formulation.

London: Oxford scientists have developed a new artificial intelligence system that can create fake videos of a person by using their still image and an audio clip. The system works by first identifying facial features using face-recognition algorithms. As the audio clip plays, the system then manipulates the mouth of the person in the still image so that it looks as if they are speaking. Although the results are not absolutely perfect, researchers believe that the software could soon make realistically fake videos only a single click away. Joon Son Chung from the University of Oxford, UK, said "The application we're thinking of is redubbing a video into another language."

Last week taurob, together with research partner TU Darmstadt, was announced the winner of the ARGOS Challenge, powered by Oil & Gas giant Total S.A. In a three-year competition, five international teams competed to develop a robot for routine-, inspection- and emergency operations on oil & gas sites. Frequently, gas leaks on oil drilling rigs can cause an increased risk to safety and the environment. The acronym ARGOS stands for Autonomous Robot for Gas and Oil Sites, which suggests that the robot independently performs assigned tasks. If necessary, an operator can intervene at any time via a satellite-based connection from land and take control of the robot.

Back in 2007, when London was booming as the financial capital of the world, a new field called "algorithmic trading" was emerging. In essence, it is about leveraging Artificial Intelligence to place bets on financials markets faster than any human can. Like most PhD students doing AI, I was working with banks to help them build their trading algorithms, which back then represented about 3% of their activity. Fast forward to 2017, and this type of trading represents over 90% in some cases, almost completely replacing human traders in big banks. One of those victims turned out to be my own dad, a trader who worked passionately for over 40 years.

IBM Watson Group opened a $200 million Internet of Things (IoT) headquarters today in Munich, Germany, what the company is calling its biggest investment in Europe in more than two decades. There more than 1,000 IBM engineers and designers will work with IBM clients and partners in what IBM Watson general manager Harriet Green called a "collaboratory." Watson works with 6,000 clients worldwide, according to the company. "This is more than a ribbon cutting or a ceremony. This is an industry moment. We think it is a turning point because at IBM we have always believed that there is only one way to fill the potential of this truly transformational technology, and that is together," Green said today at Genius of Things Summit, a gathering of press, the IoT team, and more than 400 business partners.

A principled approach to understand network structures is to formulate generative models. Given a collection of models, however, an outstanding key task is to determine which one provides a more accurate description of the network at hand, discounting statistical fluctuations. This problem can be approached using two principled criteria that at first may seem equivalent: selecting the most plausible model in terms of its posterior probability; or selecting the model with the highest predictive performance in terms of identifying missing links. Here we show that while these two approaches yield consistent results in most of cases, there are also notable instances where they do not, that is, where the most plausible model is not the most predictive. We show that in the latter case the improvement of predictive performance can in fact lead to overfitting both in artificial and empirical settings. Furthermore, we show that, in general, the predictive performance is higher when we average over collections of models that are individually less plausible, than when we consider only the single most plausible model.