In accessibility tests for digital preservation, over time we experience drifts of localized and labelled content in statistical models of evolving semantics represented as a vector field. This articulates the need to detect, measure, interpret and model outcomes of knowledge dynamics. To this end we employ a high-performance machine learning algorithm for the training of extremely large emergent self-organizing maps for exploratory data analysis. The working hypothesis we present here is that the dynamics of semantic drifts can be modeled on a relaxed version of Newtonian mechanics called social mechanics. By using term distances as a measure of semantic relatedness vs. their PageRank values indicating social importance and applied as variable `term mass', gravitation as a metaphor to express changes in the semantic content of a vector field lends a new perspective for experimentation. From `term gravitation' over time, one can compute its generating potential whose fluctuations manifest modifications in pairwise term similarity vs. social importance, thereby updating Osgood's semantic differential. The dataset examined is the public catalog metadata of Tate Galleries, London.

We study indeterminacies in realization of ornaments and how they can be incorporated in a stochastic performance model applicable for music information processing such as score-performance matching. We point out the importance of temporal information, and propose a hidden Markov model which describes it explicitly and represents ornaments with several state types. Following a review of the indeterminacies, they are carefully incorporated into the model through its topology and parameters, and the state construction for quite general polyphonic scores is explained in detail. By analyzing piano performance data, we find significant overlaps in inter-onset-interval distributions of chordal notes, ornaments, and inter-chord events, and the data is used to determine details of the model. The model is applied for score following and offline score-performance matching, yielding highly accurate matching for performances with many ornaments and relatively frequent errors, repeats, and skips.

We develop a unified and systematic framework for performing online nonnegative matrix factorization under a wide variety of important divergences. The online nature of our algorithm makes it particularly amenable to large-scale data. We prove that the sequence of learned dictionaries converges almost surely to the set of critical points of the expected loss function. We do so by leveraging the theory of stochastic approximations and projected dynamical systems. This result substantially generalizes the previous results obtained only for the squared-$\ell_2$ loss. Moreover, the novel techniques involved in our analysis open new avenues for analyzing similar matrix factorization problems. The computational efficiency and the quality of the learned dictionary of our algorithm are verified empirically on both synthetic and real datasets. In particular, on the tasks of topic learning, shadow removal and image denoising, our algorithm achieves superior trade-offs between the quality of learned dictionary and running time over the batch and other online NMF algorithms.

The Republican National Convention already seems to have solidified its position as the most dramatic political event in recent memory -- and yes, that's saying a lot. From accusations of plagiarism to a speaker lineup that reads more like a who's who of reality television than the American political system, there's a lot to cover at the 2016 RNC. And at both the Washington Post and Buzzfeed, some of that coverage is coming from bots. While the two publications may seem like rather dissimilar outlets with disparate readerships, both seem to have recognized the value in utilizing bot technology to cover the highly buzzed-about convention. The Washington Post's bot will work with Twitter and Double Robotics, to provide a live stream of the goings-on in Cleveland by way of Periscope. Moreover, readers (or rather, viewers) will be able to ask questions about the convention through a Periscope chat.

Last week, I was in San Francisco's Mission District, betwixt testing Prisma filters on my photos, and enjoying the fine cuisine, when I noticed an actual painting of the Golden Gate Bridge on the wall. "It's like Prisma in real life," I reacted. In today's technology-fueled hyper-sensualized world, one of the many symptoms is a blurring of the line between art and design. Design is function-oriented, and though some readily fetishize consumer electronics as art objects, the bounds of design are neatly wrapped in utility. Art, on the other hand, is characterized by an effort that redefines the confines of knowledge with a particular emphasis on questioning the boundaries of emotions, politics, and society.

HAVANA – Cuba may be one of the world's least connected countries but that is not stopping the Japanese subculture of animated movies, manga comics and video games from spreading feverishly among its youth. More than 1,000 otaku (geeks), or fans of such fantasy worlds, descended on Havana last week for the country's third otaku festival, defying the sweltering heat to sport the costumes of their favorite characters. Some performed scenes from animated movies on stage, while others belted out songs in Japanese with Spanish subtitles projected in the background. Others did role-playing dance choreographies. A prize was awarded for best cosplay, or role-playing in costume, and for best manga drawing.

The Queensland government has announced a AU 1 million investment in remotely piloted aircraft (RPA) technology, expected to benefit the LNG, agriculture, mining, energy, telecommunications, search and rescue, and environmental management industries. In addition to the cash injection, the state government has partnered with aerospace giant The Boeing Company, in conjunction with Boeing subsidiary Insitu Pacific, Shell's QGC project, and Telstra to further the drone research. Local small to medium-sized businesses specialising in related technology such as aerial photography, surveying, product development, and training for drone operators will also be consulted as part of the venture. "The project aims to capitalise on the capabilities inherent in drones to carry out remote-monitoring and inspection of key infrastructure and data analysis to allow for better decision-making," Queensland Premier Annastacia Palaszczuk said in a statement. In addition to creating 500 new jobs, Palaszczuk said she expects the technologies to be developed will include an improved airspace situational awareness prototype system that will enable the safe operation of RPAs over a broad area, as well as tools for enhanced data analytics.

This promoted post is an advertisement generated with our self-serve advertisement tool. Use of this tool is open to all members of reddit.com, This is a new ad format that we are currently testing. We often try new types of ads in a limited capacity. If you have feedback, please let us know in the ads subreddit.

I'm going to try to make the case that note detection in music is essentially image recognition with a few small differences and I'll describe some techniques I used to modify neural networks from computer vision to produce sheet music transcriptions of (polyphonic) music that are actually quite playable. Convolutional neural networks (CNNs) have produced the most accurate results in computer vision for several years. In a typical CNN you start with an image as a 3 dimensional array (width, height, and 3 color channels) and then pass that data through several layers of convolutions, max pooling, and some kind of non-linearity, like a ReLU. Backpropagation is used to iteratively update the convolution parameters from a set of labeled training data (pairs of input and desired output). This process builds up a sophisticated function composed of many simpler functions, primarily convolutions.

In the late 1700s, machinists started making music boxes: intricate little mechanisms that could play harmonies and melodies by themselves. Some incorporated bells, drums, organs, even violins, all coordinated by a rotating cylinder. The more ambitious examples were Lilliputian orchestras, such as the Panharmonicon, invented in Vienna in 1805, or the mass-produced Orchestrion that came along in Dresden in 1851. But the technology had limitations. To make a convincing violin sound, one had to create a little simulacrum of a violin -- quite an engineering feat. How to replicate a trombone? The artisans assumed that an entire instrument had to be copied in order to capture its distinctive tone. The metal, the wood, the reed, the shape, the exact resonance, all of it had to be mimicked. How else were you going to create an orchestral sound?