NYU Data Science newsletter – June 25, 2015

GROUP CURATION: N/A

Plottable.js is a library of chart components for creating flexible, custom charts for websites. It is built on top of D3.js and provides higher-level pieces, like plots, gridlines, and axes. As such, it’s easier to quickly build charts than with D3, and the charts are much more flexible than standard-template charts provided by charting libraries. You can think of Plottable as a “D3 for Charts” – it is not a charting library but a library of chart components.

… Automatic data collection sites like last.fm (when not abused) can really be amazing resources for data scientists, and linking data to particular users is important for it to be meaningful. But do we really need public-facing “usernames”? What if our profiles were just User1823471625987 and that was a hash that rotated every 24 hours, or every week. It’s enough time to share a link on reddit, but no one’s going to find all your data years down the line. And for real life friends, how often are they going to be checking your profile anyway? Why not send them a screen grab or just tell them the highlights?

The comparison between MongoDB, the poster child of NoSQL, and MySQL has been raging for a while now. It is important that you know the difference between the two as this will assist you in making an informed decision.

… Many patients end up getting set up with the wrong doctor the first time out and end up having to be referred again. This can detract from the quality and timeliness of the care the patient receives, and it costs the health care system money when that first exam has to be repeated by a second doctor.

One company, Kyruus, is doing a deep dive into the data to make health providers far smarter about matching patients with physicians.

As Spotify faces increasing pressure from rival streaming music services like Apple Music, Google Play Music, it’s turning to data science to help distinguish it from the pack.

Spotify said on Wednesday that it had acquired a small data analytics startup, Seed Scientific, which previously crunched data for Spotify competitor Beats Music, which Apple gobbled up last year for $3 billion. Apple has since absorbed Beats into its Apple Music streaming service.

Amazon is rolling out a big change to its customer reviews system in the US, introducing a new machine-learning platform it developed in-house to surface newer and more helpful reviews.

“The system will learn what reviews are most helpful to customers…and it improves over time,” Amazon spokeswoman Julie Law said in an interview. “It’s all meant to make customer reviews more useful.”

Deep networks have recently enjoyed enormous success when applied to recognition and classification problems in computer vision, but their use in graphics problems has been limited. In this work, we present a novel deep architecture that performs new view synthesis directly from pixels, trained from a large number of posed image sets. In contrast to traditional approaches which consist of multiple complex stages of processing, each of which require careful tuning and can fail in unexpected ways, our system is trained end-to-end. The pixels from neighboring views of a scene are presented to the network which then directly produces the pixels of the unseen view. The benefits of our approach include generality (we only require posed image sets and can easily apply our method to different domains), and high quality results on traditionally difficult scenes. We believe this is due to the end-to-end nature of our system which is able to plausibly generate pixels according to color, depth, and texture priors learnt automatically from the training data. To verify our method we show that it can convincingly reproduce known test views from nearby imagery. Additionally we show images rendered from novel viewpoints. To our knowledge, our work is the first to apply deep learning to the problem of new view synthesis from sets of real-world, natural imagery.

It’s hard to think of a more important source of information in the world than quality journalism. At its best, news communicates truth to power, keeps societies free and open, and leads to more informed decision-making by people and leaders. In the past decade, better technology and an open Internet have led to a revolution in how news is created, distributed, and consumed. And given Google’s mission to ensure quality information is accessible and useful everywhere, we want to help ensure that innovation in news leads to a more informed, more democratic world.

That’s why we’ve created the News Lab, a new effort at Google to empower innovation at the intersection of technology and media.

With the proliferation of IoT-enabled systems and devices across the physical environment, the IoT is already embedded in the enterprise, often deep inside the industrial infrastructure. Generating terabytes of data, sensor-laden objects — lighting, HVAC systems, thermostats, beacons, and others — justify their cost based on the business benefits alone. Whether for energy savings, process efficiency, worker tracking, threat detection, or other purposes, facilities engineering teams are procuring these sensors and systems to address business needs. The challenge? Buildings are getting smarter without explicit or intentional involvement from IT, creating missed opportunities to leverage these assets and their data, and to align with a broader IT strategy.

MIT’s Hacking Medicine program, which has organized medical hackathons since 2011, has launched a new spin-off: the Hacking Medicine Institute, a 501c3 nonprofit with a slightly different mission. The new organization will assess whether digital health products and services really work and, if they do, help them to prove their efficacy to consumers, doctors, and insurers.

Critical Data Studies (CDS) is a growing field of research that focuses on the unique theoretical, ethical, and epistemological challenges posed by “Big Data.” Rather than treat Big Data as a scientifically empirical, and therefore largely neutral phenomena, CDS advocates the view that data should be seen as always-already constituted within wider data assemblages. Assemblages is a concept that helps capture the multitude of ways that already-composed data structures inflect and interact with society, its organization and functioning, and the resulting impact on individuals’ daily lives. CDS questions the many assumptions about data that permeate contemporary literature on information and society by locating instances where data may be naively taken to denote objective and transparent informational entities.

Proposal Deadline: July 10, 2015 … Proposals of 1000 words are invited for consideration and inclusion in the Special Theme for an Original Research Article, Commentary, or essay in the Early Career Research Forum section.