Week Ending 09.15.19
RESEARCH WATCH: 09.15.19
Over the past week, 999 new papers were published in "Computer Science".
The paper discussed most in the news over the past week was "Differentially Private SQL with Bounded User Contribution" by Royce J Wilson et al (Sep 2019), which was referenced 16 times, including in the article Google hopes to protect users with open source differential privacy library in Tech Republic. The paper author, Damien Desfontaines (Google privacy software engineer), was quoted saying "OK so why am I so excited about this release? So many reasons. First, the code is the same one we use internally. It powers massive-scale tools and major use cases".
Leading researcher Kyunghyun Cho (New York University) published "Finding Generalizable Evidence by Learning to Convince Q&A Models", which has 0 shares on Twitter so far.
This week was very active for "Computer Science - Artificial Intelligence", with 164 new papers.
The paper discussed most in the news over the past week was by a team at Allen Institute for Artificial Intelligence: "From F to A on the N.Y. Regents Science Exams: An Overview of the Aristo Project" by Peter Clark et al (Sep 2019), which was referenced 9 times, including in the article AI Can Pass Standardized Tests—But It Would Fail Preschool in Wired News. The paper got social media traction with 134 shares. A user, @peterjansen_ai, tweeted "Here is the paper describing AI2's system that achieves 90% accuracy on the 8th grade standardized science exams. It's truly incredible to see how much of this comes from contextualized embeddings -- I never thought they would do that well on this complex inference task".
Leading researcher Kyunghyun Cho (New York University)
The paper shared the most on social media this week is by a team at University of Washington Seattle: "Meta-Learning with Implicit Gradients" by Aravind Rajeswaran et al (Sep 2019) with 177 shares. @hillbig (Daisuke Okanohara) tweeted "Meta-learning requires inner-loop optimization for each task, and implicit differentiation can the gradient directly from the solution. (iMAML) Similar idea is also proposed in hierarchical Bayesian meta-learning setting".
This week was active for "Computer Science - Computer Vision and Pattern Recognition", with 221 new papers.
The paper discussed most in the news over the past week was "Progressive Face Super-Resolution via Attention to Facial Landmark" by Deokyun Kim et al (Aug 2019), which was referenced 13 times, including in the article Facebook Leaks Data of 419M Users; NeurIPS 2019 Accepted Papers Announced; AI Passes 8th Grade Science Test in SyncedReview.com. The paper also got the most social media traction with 1068 shares. A Twitter user, @JohnAndrewsX, posted "So if AI can do this today, imagine what happens when implement AI-drivatars in Forza games one day >> True Unbeatable Mode 🤘😈🤘 #ForzaHorizon #ForzaMotorsport".
The paper shared the most on social media this week is by a team at Google: "A deep learning system for differential diagnosis of skin diseases" by Yuan Liu et al (Sep 2019) with 215 shares. The researchers developed a deep learning system (DLS) to provide a differential diagnosis of skin conditions for clinical cases (skin photographs and associated medical histories). @Med_ETH (Jörg Goldhahn, MD, MAS) tweeted "What will be possible in the next years when our current medical students start their jobs? #meded".
Over the past week, 18 new papers were published in "Computer Science - Computers and Society".
The paper discussed most in the news over the past week was "Auditing News Curation Systems: A Case Study Examining Algorithmic and Editorial Logic in Apple News" by Jack Bandy et al (Aug 2019), which was referenced 3 times, including in the article Apple News is excluding local newsrooms from its coveted traffic bump in Columbia Journalism Review. The paper author, Nicholas Diakopoulos, was quoted saying "Automating the News: How Algorithms are Rewriting the Media". The paper got social media traction with 10 shares. A Twitter user, @jackbandy, commented "Link 3/4: a preprint of the full paper for the hardcore nerds".
This week was active for "Computer Science - Human-Computer Interaction", with 30 new papers.
The paper discussed most in the news over the past week was "Auditing News Curation Systems: A Case Study Examining Algorithmic and Editorial Logic in Apple News" by Jack Bandy et al (Aug 2019)
This week was very active for "Computer Science - Learning", with 393 new papers.
The paper discussed most in the news over the past week was by a team at Google: "Personalizing ASR for Dysarthric and Accented Speech with Limited Data" by Joel Shor et al (Jul 2019), which was referenced 16 times, including in the article Google at Interspeech 2019 in CHROME RELEASES. The paper got social media traction with 27 shares. The researchers present and evaluate finetuning techniques to improve ASR for users with non - standard speech. A Twitter user, @DataScienceNIG, said "Speak better with Artificial Intelligence - ASR from & for slurred speech & those with accents - a speech2text transcription for people with speaking impairments &71% of the improvement comes from only 5 mins of training data. Read at".
Leading researcher Kyunghyun Cho (New York University) published "Countering Language Drift via Visual Grounding".
The paper shared the most on social media this week is by a team at University of Washington Seattle: "Meta-Learning with Implicit Gradients" by Aravind Rajeswaran et al (Sep 2019)
Over the past week, 19 new papers were published in "Computer Science - Multiagent Systems".
Leading researcher Kyunghyun Cho (New York University) came out with "Finding Generalizable Evidence by Learning to Convince Q&A Models".
Over the past week, 30 new papers were published in "Computer Science - Neural and Evolutionary Computing".
The paper discussed most in the news over the past week was "Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches" by Maurizio Ferrari Dacrema et al (Jul 2019), which was referenced 1 time, including in the article RecoTour II: neural recommendation algorithms in Towards Data Science. The paper also got the most social media traction with 2807 shares. The investigators report the results of a systematic analysis of algorithmic proposals for top - n recommendation tasks. On Twitter, @ekshakhs commented ""Consider 18 algorithms that were presented at top-level research conferences in the last years. Only 7 of them could be reproduced with reasonable effort... 6 of them can often be outperformed with comparably simple heuristic methods". Systemic problem with bleeding research!".
This week was very active for "Computer Science - Robotics", with 91 new papers.