Week Ending 04.07.19

 

RESEARCH WATCH: 04.07.19

 
ai-research.png

Over the past week, 278 new papers were published in "Computer Science".

Over the past week, 74 new papers were published in "Computer Science - Artificial Intelligence".

  • The paper discussed most in the news over the past week was by a team at University of Cambridge: "Hearing your touch: A new acoustic side channel on smartphones" by Ilia Shumailov et al (Mar 2019), which was referenced 3 times, including in the article Recovering Smartphone Typing from Microphone Sounds in FinTechLog.com. The paper got social media traction with 47 shares. On Twitter, @ITnextLevel observed "Very impressive! Detect what was done in a smartphone via the microphone", while @shoemaneu commented "Your Smartphone PIN is not as secure as you think. Hearing what you tap".

  • Leading researcher Pieter Abbeel (University of California, Berkeley) came out with "Guided Meta-Policy Search".

  • The paper shared the most on social media this week is "Reducing BERT Pre-Training Time from 3 Days to 76 Minutes" by Yang You et al (Apr 2019) with 146 shares. @bsaeta (Brennan Saeta) tweeted "Cloud TPU Pods accelerate the pace of research. Using the latest versions of , Google researcher and TFEngineers are able to reduce BERT pretraining time from 3 days down to 76 minutes using a new optimizer. Check out the details".

This week was active for "Computer Science - Computer Vision and Pattern Recognition", with 295 new papers.

  • The paper discussed most in the news over the past week was "Predictive Inequity in Object Detection" by Benjamin Wilson et al (Feb 2019), which was referenced 144 times, including in the article Face recognition researcher fights Amazon over biased AI in Minneapolis Star Tribune. The paper author, Jamie Morgenstern (University of Pennsylvania), was quoted saying "The main takeaway from our work is that vision systems that share common structures to the ones we tested should be looked at more closely". The paper got social media traction with 167 shares. The researchers investigate whether state - of - the - art object detection systems have equitable predictive performance on pedestrians with different skin tones. A Twitter user, @defcon_5, commented "This is what systemic racism looks like.🚨🚨🚨 Black people may be at a greater risk of getting hit by self-driving cars because today's object-detection models exhibit higher precision on lighter skin tones #whitesupremacy".

  • Leading researcher Devi Parikh (Georgia Institute of Technology) published "Habitat: A Platform for Embodied AI Research", which had 23 shares over the past 3 days.

  • The paper shared the most on social media this week is "HoloGAN: Unsupervised learning of 3D representations from natural images" by Thu Nguyen-Phuoc et al (Apr 2019) with 508 shares. @richardmatthias (Richard Matthias) tweeted "Nice. I was thinking about something like this a few months ago (in response to the school bus/snow plow thing). I love that there's so much ML research now that you can pretty much guarantee that whatever idea you have, someone somewhere is already working on it :)".

Over the past week, 24 new papers were published in "Computer Science - Computers and Society".

  • The paper discussed most in the news over the past week was "Discrimination through optimization: How Facebooks ad delivery can lead to skewed outcomes" by Muhammad Ali et al (Apr 2019), which was referenced 45 times, including in the article Facebook's ad targeting discriminates by race and gender, even when it's not told to, study suggestsin Business Insider. The paper author, Alan Mislove (Computer science professor at Northeastern University), was quoted saying "Ultimately we don’t know what Facebook is doing". The paper also got the most social media traction with 1258 shares. The researchers demonstrate that such skewed delivery occurs on Facebook, due to market and financial optimization effects as well as the platforms own predictions about the relevance of ads to different groups of users. A user, @TimKarr, tweeted "April 4, 2019: Academic paper published analyzing FB's algorithms to find they're built using historically discriminatory data. The algorithms deliver results biased against people based on race & gender, & perpetuate discrimination in advertising".

This week was active for "Computer Science - Human-Computer Interaction", with 25 new papers.

  • The paper discussed most in the news over the past week was "Modeling Mobile Interface Tappability Using Crowdsourcing and Deep Learning" by Amanda Swearngin et al (Feb 2019), which was referenced 3 times, including in the article Google’s AI-powered app usability testing promises human-level accuracy in Venturebeat. The paper author, Yang Li (Iowa State University), was quoted saying "Tapping is the most commonly used gesture on mobile interfaces, and is used to trigger all kinds of actions ranging from launching an app to entering text … [but] predicting tappability is merely one example of what we can do with machine learning to solve usability issues in user interfaces". The paper got social media traction with 9 shares. The researchers present an approach for modeling tappability of mobile interfaces at scale.

  • The paper shared the most on social media this week is by a team at Stanford University: "HYPE: Human eYe Perceptual Evaluation of Generative Models" by Sharon Zhou et al (Apr 2019) with 55 shares. @charleskfisher (Charles Fisher) tweeted "That is "How should we judge generative models *of images*?" After all, it is possible to create generative models of things that are not images. 😉".

This week was active for "Computer Science - Learning", with 189 new papers.

  • The paper discussed most in the news over the past week was "Predictive Inequity in Object Detection" by Benjamin Wilson et al (Feb 2019), which was referenced 143 times, including in the article Face recognition researcher fights Amazon over biased AI in Minneapolis Star Tribune. The paper author, Jamie Morgenstern (University of Pennsylvania), was quoted saying "The main takeaway from our work is that vision systems that share common structures to the ones we tested should be looked at more closely". The paper got social media traction with 167 shares. The authors investigate whether state - of - the - art object detection systems have equitable predictive performance on pedestrians with different skin tones. A user, @defcon_5, tweeted "This is what systemic racism looks like.🚨🚨🚨 Black people may be at a greater risk of getting hit by self-driving cars because today's object-detection models exhibit higher precision on lighter skin tones #whitesupremacy".

  • Leading researcher Kyunghyun Cho (New York University) published "Molecular geometry prediction using a deep generative graph neural network" The authors propose a conditional deep generative graph neural network that learns an energy function from data by directly learning to generate molecular conformations given a molecular graph. @debasishg tweeted "Molecular geometry prediction using a deep generative graph neural network .. 👈 very cool stuff".

  • The paper shared the most on social media this week is by a team at DeepMind: "Analysing Mathematical Reasoning Abilities of Neural Models" by David Saxton et al (Apr 2019) with 1051 shares. The researchers present a new challenge for the evaluation (and eventually the design) of neural architectures and similar system, developing a task suite of mathematics problems involving sequential questions and answers in a free - form textual input/output format. @pkedrosky (Paul Kedrosky) tweeted "DeepMind taught itself basic math, but got 14 out of 40 on a common exam taken by 16-year-olds in the UK. Fun confusion: • It could successfully add 1+1+1+1+1+1, but failed when an extra 1 was added • It knew 17*4 was 68, but thought 17*4. was 69".

  • The most influential Twitter user discussing papers is Fiona/ @RealScientists who shared "Positron Annihilation in the Nuclear Outflows of the Milky Way" by Fiona H. Pantheret al (Oct 2017) and said: "First paper: My second paper reviews everything we know about how positrons float around the Galaxy!".

Over the past week, seven new papers were published in "Computer Science - Multiagent Systems".

Over the past week, 19 new papers were published in "Computer Science - Neural and Evolutionary Computing".

  • The paper discussed most in the news over the past week was by a team at Eindhoven University of Technology: "Learning with Delayed Synaptic Plasticity" by Anil Yaman et al (Mar 2019), which was referenced 1 time, including in the article A bio-inspired approach to enhance learning in ANNs in PhysOrg.com. The paper author, Anil Yaman(Eindhoven University of Technology), was quoted saying "We are mainly interested in understanding the emergent behavior and learning dynamics of artificial neural networks , and developing a coherent model to explain how synaptic plasticity occurs in different learning scenarios". The paper was shared 2 times in social media. A user, @Eschersand, tweeted "Modelling plasticity, or rather evolving it, has been a long-standing goal in Neuro-evolution (NE), a research field that aims to design artificial neural networks (ANNs) using evolutionary computing approaches. We propose neuron activation traces (NATs) and evolve (DSP) rules".

This week was active for "Computer Science - Robotics", with 54 new papers.

EYE ON A.I. GETS READERS UP TO DATE ON THE LATEST FUNDING NEWS AND RELATED ISSUES. SUBSCRIBE FOR THE WEEKLY NEWSLETTER.

 
Artificial Intelligence, Research WatchCraig SmithComment