The Kalman filter (KF) is used in a variety of applications for computing the posterior distribution of latent states in a state space model. The model requires a linear relationship between states and observations. Extensions to the Kalman filter have been proposed that incorporate linear approximations to nonlinear models, such as the extended Kalman filter (EKF) and the unscented Kalman filter (UKF). However, we argue that in cases where the dimensionality of observed variables greatly exceeds the dimensionality of state variables, a model for $p(\text{state}|\text{observation})$ proves both easier to learn and more accurate for latent space estimation. We derive and validate what we call the discriminative Kalman filter (DKF): a closed-form discriminative version of Bayesian filtering that readily incorporates off-the-shelf discriminative learning techniques. Further, we demonstrate that given mild assumptions, highly non-linear models for $p(\text{state}|\text{observation})$ can be specified. We motivate and validate on synthetic datasets and in neural decoding from non-human primates, showing substantial increases in decoding performance versus the standard Kalman filter.

In image deconvolution problems, the diagonalization of the underlying operators by means of the FFT usually yields very large speedups. When there are incomplete observations (e.g., in the case of unknown boundaries), standard deconvolution techniques normally involve non-diagonalizable operators, resulting in rather slow methods, or, otherwise, use inexact convolution models, resulting in the occurrence of artifacts in the enhanced images. In this paper, we propose a new deconvolution framework for images with incomplete observations that allows us to work with diagonalized convolution operators, and therefore is very fast. We iteratively alternate the estimation of the unknown pixels and of the deconvolved image, using, e.g., an FFT-based deconvolution method. This framework is an efficient, high-quality alternative to existing methods of dealing with the image boundaries, such as edge tapering. It can be used with any fast deconvolution method. We give an example in which a state-of-the-art method that assumes periodic boundary conditions is extended, through the use of this framework, to unknown boundary conditions. Furthermore, we propose a specific implementation of this framework, based on the alternating direction method of multipliers (ADMM). We provide a proof of convergence for the resulting algorithm, which can be seen as a "partial" ADMM, in which not all variables are dualized. We report experimental comparisons with other primal-dual methods, where the proposed one performed at the level of the state of the art. Four different kinds of applications were tested in the experiments: deconvolution, deconvolution with inpainting, superresolution, and demosaicing, all with unknown boundaries.

We study the problem of determining the one-dimensional structure that best represents a given data set. More precisely, we take a variational approach to approximating a given measure (data) by curves. We consider an objective functional whose minimizers are a regularization of principal curves and introduce a new functional which allows for multiple curves. We prove existence of minimizers and investigate their properties. While both of the functionals used are non-convex, we show that enlarging the configuration space to allow for multiple curves leads to a simpler energy landscape with fewer undesirable (high-energy) local minima. We provide an efficient algorithm for approximating minimizers of the functional and demonstrate its performance on real and synthetic data. The numerical examples illustrate the effectiveness of the proposed approach in the presence of substantial noise, and the viability of the algorithm for high-dimensional data.

The support recovery problem consists of determining a sparse subset of a set of variables that is relevant in generating a set of observations, and arises in a diverse range of settings such as compressive sensing, and subset selection in regression, and group testing. In this paper, we take a unified approach to support recovery problems, considering general probabilistic models relating a sparse data vector to an observation vector. We study the information-theoretic limits of both exact and partial support recovery, taking a novel approach motivated by thresholding techniques in channel coding. We provide general achievability and converse bounds characterizing the trade-off between the error probability and number of measurements, and we specialize these to the linear, 1-bit, and group testing models. In several cases, our bounds not only provide matching scaling laws in the necessary and sufficient number of measurements, but also sharp thresholds with matching constant factors. Our approach has several advantages over previous approaches: For the achievability part, we obtain sharp thresholds under broader scalings of the sparsity level and other parameters (e.g., signal-to-noise ratio) compared to several previous works, and for the converse part, we not only provide conditions under which the error probability fails to vanish, but also conditions under which it tends to one.

Load disaggregation based on aided linear integer programming (ALIP) is proposed. We start with a conventional linear integer programming (IP) based disaggregation and enhance it in several ways. The enhancements include additional constraints, correction based on a state diagram, median filtering, and linear programming-based refinement. With the aid of these enhancements, the performance of IP-based disaggregation is significantly improved. The proposed ALIP system relies only on the instantaneous load samples instead of waveform signatures, and hence does not crucially depend on high sampling frequency. Experimental results show that the proposed ALIP system performs better than the conventional IP-based load disaggregation system.

The market for artificial intelligence in the US is projected to grow at a CAGR of around 75% during 2016-2021. United States is the largest market for artificial intelligence solutions, globally. Continuous research and development in healthcare, autonomous vehicles, security and access control, cyber security, etc., are expected to fuel growth in the United States artificial intelligence market. Rising penetration of smart wearables, burgeoning head-up display screen market in luxury car segment and growing venture capital investments are also propelling adoption of artificial intelligence solutions in the country. United States government is also playing a proactive role in the country's artificial intelligence market by organizing workshops in association with leading solution providers.

Tech giant Microsoft has acquired artificial intelligence (AI)-based scheduling service Genee that simplifies the scheduling and rescheduling of large group meetings for companies. Microsoft will integrate the AI technology into Office 365 before shutting down Genee. Co-founders Ben Cheung and Charles Lee, who plan to join Microsoft, started Genee in 2014 to simplify the time-consuming task of scheduling (and rescheduling) meetings. "It is especially useful for large groups and for when you don't have access to someone's calendar," Jha added. Genee uses natural language processing and optimised decision-making algorithms so that interacting with a virtual assistant is just like interacting with a human one.

A groundbreaking study to investigate how dog brains process speech has revealed canines care about both what we say and how we say it. It discovered that dogs, like people, use the left hemisphere to process words, and the right hemisphere brain region to process intonation. It found praise activates dog's reward centre only when both words and intonation match, according to the new study in Science. Trained dogs around the fMRI scanner used in the study: Dogs, like people, use the left hemisphere to process words, and the right hemisphere brain region to process intonation, according to the new study in Science. The brain activation images showed that dogs prefer to use their left hemisphere to process meaningful but not meaningless words.

For anyone who's heard the "Oompa Loompa" song or Frankenstein pronounced as "Fronkensteen," news of actor Gene Wilder's death on Monday cuts deeply. Wilder died from complications of Alzheimer's disease at age 83, according to his nephew Jordan Walker-Pearlman. Wilder passed away in his hometown of Stamford, Connecticut, late Sunday, reported the Associated Press. He had been diagnosed with non-Hodgkin's lymphoma, a cancer that originates in the lymphatic system, in 1989. Wilder was born Jerome Silberman in Milwaukee and began studying acting at age 12.

The mystery of how Parkinson's disease progresses could be cracked thanks to researchers at the Australian National University (ANU) and machine learning. Deborah Apthorp of the ANU Research School of Psychology has won funding for a study that will track early symptoms with the aim of finding possible indicators of progression, using machine learning. This research has received 138,930 over 5 years from the Perpetual Impact Philanthropy Grant. As it stands, the type of Parkinson's a diagnosed patient has or how quickly it will progress is hard to determine. Apthorp noted in an ANU report that some individuals can be fine for quite a while while others can experience a more rapid progression.