NYU Data Science newsletter – September 29, 2016

GROUP CURATION: N/A

A couple of weeks ago, I posted about the origins of the Data-Driven Discovery Initiative, and outlined what we are funding and why. In this post, I dive into the structure of the initiative, our anticipated gains from a mid-term evaluation that’s just beginning, and how that information will feed into a new strategic plan for 2018 and beyond.

Let’s face it: artificial intelligence is scary. After decades of dystopian science fiction novels and movies where sentient machines end up turning on humanity, we can’t help but worry as real world AI continues to improve at such a rapid rate. Sure, that danger is probably decades away if it’s even a real danger at all. But there are many more immediate concerns. Will automated robots cost us jobs? Will online face recognition destroy our privacy? Will self-driving cars mess with moral decision making?

The good news is that many of the tech giants behind the new wave of AI are well aware that it scares people—and that these fears must be addressed. That’s why Amazon, Facebook, Google’s DeepMind division, IBM, and Microsoft have founded a new organization called the Partnership on Artificial Intelligence to Benefit People and Society.

“Every new technology brings transformation, and transformation sometimes also causes fear in people who don’t understand the transformation,” Facebook’s director of AI Yann LeCun said this morning during a press briefing dedicated to the new project. “One of the purposes of this group is really to explain and communicate the capabilities of AI, specifically the dangers and the basic ethical questions.”

Over the past four years, readers have doubtlessly noticed quantum leaps in the quality of a wide range of everyday technologies.

Most obviously, the speech-recognition functions on our smartphones work much better than they used to. When we use a voice command to call our spouses, we reach them now. We aren’t connected to Amtrak or an angry ex.

In fact, we are increasingly interacting with our computers by just talking to them, whether it’s Amazon’s Alexa, Apple’s Siri, Microsoft’s Cortana, or the many voice-responsive features of Google. Chinese search giant Baidu says customers have tripled their use of its speech interfaces in the past 18 months.

Skymind, a startup that promotes the use of the Deeplearning4j open-source software for deep learning — a type of artificial intelligence (A.I.) — is announcing today that it has raised a $3 million seed round.

There are many open-source frameworks for deep learning, which generally involves training artificial neural nets on lots of data, such as labeled photos, and then getting them to make inferences about new data. Deeplearning4j is primarily written in Java and is meant to be integrated into companies’ existing architectures.

Recent neural network sequence models with softmax classifiers have achieved their best language modeling performance only with very large hidden states and large vocabularies. Even then they struggle to predict rare or unseen words even if the context makes the prediction unambiguous. We introduce the pointer sentinel mixture architecture for neural sequence models which has the ability to either reproduce a word from the recent context or produce a word from a standard softmax classifier. Our pointer sentinel-LSTM model achieves state of the art language modeling performance on the Penn Treebank (70.9 perplexity) while using far fewer parameters than a standard softmax LSTM. In order to evaluate how well language models can exploit longer contexts and deal with more realistic vocabularies and larger corpora we also introduce the freely available WikiText corpus.

As you browse the internet, online advertisers track nearly every site you visit, amassing a trove of information on your habits and preferences. When you visit a news site, they might see you’re a fan of basketball, opera and mystery novels, and accordingly select ads tailored to your tastes. Advertisers use this information to create highly personalized experiences, but they typically don’t know exactly who you are. They observe only your digital trail, not your identity itself, and so you might feel that you’ve retained a degree of anonymity.

In new work with Ansh Shukla, Sharad Goel and Arvind Narayanan, we show that these anonymous web browsing records can in fact often be tied back to real-world identities.

Israel-based Beyond Verbal, which makes voice recognition software to analyze human emotion, is now moving its work into health and wellness with the launch of the Beyond mHealth Research platform. The goal is to identify physiological markers through the voice that may indicate various health related issues.

Researchers can now integrate Beyond Verbal’s API into any type of application and correlate it other life events, leveraging the existing AI tools as well as the expertise of its data science team to gain new insight into unique vocal biomarkers. When a new biomarker is discovered, the research platform will guide partners on the path to scaling up their findings. The platform also offers a HIPAA compliant app that collects participants’ data and a secure web interface to facilitate data sharing.

AI NexusLab is a four-month program run by the NYU Future Labs to support AI companies’ going from ideation and MVP and product-market fit. The AI NexusLab will recruit the top AI startups from across the world to come to NYC for a four-month program. Companies will receive $100k to join the lab, and will gain access to two full-time technical experts, a network of mentors including NYU AI faculty experts, abundant resources, and a rigorous program to guide startups to market entry.

A Roadmap towards Machine Intelligence.”

“The NVIDIA® CUDA® Toolkit provides a comprehensive development environment for C and C++ developers building GPU-accelerated applications including a compiler for NVIDIA GPUs, math libraries, and tools for debugging and optimizing application performance. You’ll also find programming guides, user manuals, API reference, and other documentation.”

The tensorflow package provides access to the complete TensorFlow API from within R.

Welcome to the inaugural Software and Data Carpentry newsletter. We are launching this newsletter to keep you well informed around all of the various activities and opportunities in the community.

Palo Alto-based healthcare app and analytics platform Medable announced the launch of Axon, a SmartStudy system that enables researchers to create and deploy ResearchKit app for clinical trials or research studies on their own, without having to work with a developer. The company formally introduced the offering at Health 2.0 in Santa Clara.

“Once ReasearchKit came out, we immediately saw a huge opportunity to better understand any research study, especially if we had more data and specifically patient-generated data from a mobile device,” Dr. Michelle Longmire, CEO of Medable told MobiHealthNews.

“We announce the Google Neural Machine Translation system (GNMT). Our full research results: Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation”