Janicza Bravo makes short films about loneliness. In one, Michael Cera plays an abrasive paraplegic who can't get lucky. In another, Gaby Hoffmann plays a phone stalker for whom the description "comes on too strong" is not strong enough. Bravo's shorts employ the visual grammar of art-house cinema: over-the-shoulder shots representing a character's point of view, handheld tracking shots depicting urgent movement, lingering closeups to heighten intimacy or unease, carefully composed establishing shots with an actor in the center of the frame. In March, 2015, Bravo went to Venice, on the western edge of Los Angeles, to meet with a production company called Wevr. The name is pronounced "weaver," but it can also be thought of as a sentence, with "We" as the subject and "V.R." as the verb. As anyone who has read a tech blog within the past five years, or a sci-fi novel within the past five decades, knows, "V.R." stands for virtual reality--a loosely defined phrase that is now being applied to several related forms of visual media. You put your smartphone into a portable device like a Google Cardboard or a Samsung Gear--or you use a more powerful computer-based setup, such as the Oculus Rift or the HTC Vive--and the device engulfs your field of vision and tracks your head movement. The filmic world is no longer flat. Wherever you look, there's something to see. The producers at Wevr invited Bravo to write and direct a V.R. project. "I said no," she told me. "It sounded like a technical thing, and I'm not into technical. But then I talked to my husband, and he said, 'How often do people just hand you money in this business?' So I changed my mind." She thought about what kind of story might be told most effectively in the new medium. "The two words I kept hearing about V.R. were'empathy' and'immersion,' and I wasn't sure that being immersed in one of my dark comedies would be all that useful."

Oracle Corp. has acquired Israeli machine-learning Big Data startup Crosswise, Inc. The price of the acquisition was not officially disclosed but is believed to be 50 million according to local media. Founded in 2013, Crosswise provides an authoritative consumer device map to ad tech vendors, consumer brands, and premium publishers. The company's platform combines data science, Big Data and machine learning, to identify which PCs, phones, tablets, digital TVs and other connected devices are being used by individual consumers; by applying advanced data science and proprietary machine-learning techniques to this data, Crosswise constructs a new probabilistic Device Map matching multiple devices to individual users in an accurate, scalable and high-quality manner. According to Crosswise, the benefits in being able to provide this data is that it allows marketers and premium publishers to deliver advertising, personalization and analytics across different sorts of devices.

This semester, as a reflective practice and an opportunity to continue my hobby (filmmaking), I decided to create a weekly video log (or vlog) of what has been happening in my middle school robotics class. The following was episode 4 of the series aptly entitled'Middle School Robotics.' This video was a bit different than the others as I sort of gave an overview of the course, my philosophy behind it, and the practices that I've seen bear the most fruit in the class. Most of my project resources were taken from this website: http://ev3lessons.com/lessons.html

The problem of building an autonomous robot has traditionally been viewed as one of integration: connecting together modular components, each one designed to handle some portion of the perception and decision making process. For example, a vision system might be connected to a planner that might in turn provide commands to a low-level controller that drives the robot's motors. In this talk, I will discuss how ideas from deep learning can allow us to build robotic control mechanisms that combine both perception and control into a single system. This system can then be trained end-to-end on the task at hand. I will show how this end-to-end approach actually simplifies the perception and control problems, by allowing the perception and control mechanisms to adapt to one another and to the task.

High-quality, low-cost online courses could be used to shift schools away from being "exam factories" and help students keep pace with the threat of automation, according to a new report by the Institute of Directors. The report argues that the internet allows schools to be more flexible and adapt learning towards "a future in which more and more work is taken over by robots or computers". Related: Welcome to the robot-based workforce: will your job become automated too? "The cost savings, convenience and flexibility that online learning offers has the potential to revolutionise education provision, but only if businesses and the education sector work together to capitalise on the potential of computer-based teaching applications to support employees in their pursuit of lifelong learning," the report said. Last year the CBI's director general also called for GCSEs to be scrapped and A-levels to be augmented by vocational courses. The report also calls for new tax incentives to encourage people to return to education, and to make it easier for employers to invest in their staff.

Gaussian process models are flexible, Bayesian non-parametric approaches to regression. Properties of multivariate Gaussians mean that they can be combined linearly in the manner of additive models and via a link function (like in generalized linear models) to handle non-Gaussian data. However, the link function formalism is restrictive, link functions are always invertible and must convert a parameter of interest to a linear combination of the underlying processes. There are many likelihoods and models where a non-linear combination is more appropriate. We term these more general models Chained Gaussian Processes: the transformation of the GPs to the likelihood parameters will not generally be invertible, and that implies that linearisation would only be possible with multiple (localized) links, i.e. a chain. We develop an approximate inference procedure for Chained GPs that is scalable and applicable to any factorized likelihood. We demonstrate the approximation on a range of likelihood functions.

Kernel mean embeddings have recently attracted the attention of the machine learning community. They map measures $\mu$ from some set $M$ to functions in a reproducing kernel Hilbert space (RKHS) with kernel $k$. The RKHS distance of two mapped measures is a semi-metric $d_k$ over $M$. We study three questions. (I) For a given kernel, what sets $M$ can be embedded? (II) When is the embedding injective over $M$ (in which case $d_k$ is a metric)? (III) How does the $d_k$-induced topology compare to other topologies on $M$? The existing machine learning literature has addressed these questions in cases where $M$ is (a subset of) the finite regular Borel measures. We unify, improve and generalise those results. Our approach naturally leads to continuous and possibly even injective embeddings of (Schwartz-) distributions, i.e., generalised measures, but the reader is free to focus on measures only. In particular, we systemise and extend various (partly known) equivalences between different notions of universal, characteristic and strictly positive definite kernels, and show that on an underlying locally compact Hausdorff space, $d_k$ metrises the weak convergence of probability measures if and only if $k$ is continuous and characteristic.

This work is a further study on the Generalized Constraint Neural Network (GCNN) model [1], [2]. Two challenges are encountered in the study, that is, to embed any type of prior information and to select its imposing schemes. The work focuses on the second challenge and studies a new constraint imposing scheme for equality constraints. A new method called locally imposing function (LIF) is proposed to provide a local correction to the GCNN prediction function, which therefore falls within Locally Imposing Scheme (LIS). In comparison, the conventional Lagrange multiplier method is considered as Globally Imposing Scheme (GIS) because its added constraint term exhibits a global impact to its objective function. Two advantages are gained from LIS over GIS. First, LIS enables constraints to fire locally and explicitly in the domain only where they need on the prediction function. Second, constraints can be implemented within a network setting directly. We attempt to interpret several constraint methods graphically from a viewpoint of the locality principle. Numerical examples confirm the advantages of the proposed method. In solving boundary value problems with Dirichlet and Neumann constraints, the GCNN model with LIF is possible to achieve an exact satisfaction of the constraints.

This work studies applications and generalizations of a simple estimation technique that provides exponential concentration under heavy-tailed distributions, assuming only bounded low-order moments. We show that the technique can be used for approximate minimization of smooth and strongly convex losses, and specifically for least squares linear regression. For instance, our $d$-dimensional estimator requires just $\tilde{O}(d\log(1/\delta))$ random samples to obtain a constant factor approximation to the optimal least squares loss with probability $1-\delta$, without requiring the covariates or noise to be bounded or subgaussian. We provide further applications to sparse linear regression and low-rank covariance matrix estimation with similar allowances on the noise and covariate distributions. The core technique is a generalization of the median-of-means estimator to arbitrary metric spaces.

We consider the problem of learning a non-negative linear classifier with a $1$-norm of at most $k$, and a fixed threshold, under the hinge-loss. This problem generalizes the problem of learning a $k$-monotone disjunction. We prove that we can learn efficiently in this setting, at a rate which is linear in both $k$ and the size of the threshold, and that this is the best possible rate. We provide an efficient online learning algorithm that achieves the optimal rate, and show that in the batch case, empirical risk minimization achieves this rate as well. The rates we show are tighter than the uniform convergence rate, which grows with $k^2$.