Week Ending 4.19.2020
RESEARCH WATCH: 4.19.2020
Over the past week, 52 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 Google: "AutoML-Zero: Evolving Machine Learning Algorithms From Scratch" by Esteban Real et al (Mar 2020), which was referenced 15 times, including in the article New AI improves itself through Darwinian-style evolution in Big Think. The paper also got the most social media traction with 1444 shares. A user, @tomvarsavsky, tweeted "One of the most interesting results I've seen in ML in the last 5 years. Evolving programs using a generic search space and generic mutations leads to the discovery of not only SGD and two layer NNs but also rand init, ReLU, Grad Norm. Can someone find a hidden inductive bias?".
Leading researcher Sergey Levine (University of California, Berkeley) came out with "Thinking While Moving: Deep Reinforcement Learning with Concurrent Control" @JeffDean tweeted "Some very nice work from our robotics research team enables robots to get a faster grasp on the problem at hand by contemplating and doing at the same time. Up to 50% faster grasping!".
The paper shared the most on social media this week is by a team at University of Tübingen: "Shortcut Learning in Deep Neural Networks" by Robert Geirhos et al (Apr 2020) with 193 shares. The investigators seek to distil how many of deep learnings problem can be seen as different symptoms of the same underlying problem : shortcut learning. @hardmaru (hardmaru) tweeted "Shortcut Learning in Deep Neural Networks Humans also take shortcuts and cheat in life when we can get away with it. Interesting they mention that ways to overcome shortcut learning for artificial agents might apply to closing loopholes in human systems".
The most influential Twitter user discussing papers is Horace Dediu who shared "Generalized logistic growth modeling of the COVID-19 outbreak in 29 provinces in China and in the rest of the world" by Ke Wu et al (Mar 2020) and said: "Incidentally, the data I use is smoothed over 7 days. See also: and".
Over the past week, 184 new papers were published in "Computer Science - Computer Vision and Pattern Recognition".
The paper discussed most in the news over the past week was "Rapid AI Development Cycle for the Coronavirus (COVID-19) Pandemic: Initial Results for Automated Detection & Patient Monitoring using Deep Learning CT Image Analysis" by Ophir Gozes et al (Mar 2020), which was referenced 25 times, including in the article AI-Powered CT Imaging System Shown to Detect COVID-19 in Medical Device & Diagnostic Industry. The paper author, Adam Bernheim (Icahn School of Medicine), was quoted saying "Protocols are in place following CT scanning of a known or suspected COVID-19 patient to ensure that risk of transmission to future patients is minimized." The paper got social media traction with 57 shares. On Twitter, @primer_ai said "The website identifies emerging topics, grouping research papers by common concepts. One topic is “Deep Learning & Machine Learning” with a popular study coming from that used AI-based image analysis tools for the detection of coronavirus".
Leading researcher Jianfeng Gao (Microsoft) published "Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks" The authors While existing methods simply concatenate image region features and text features as input to the model to be pre-trained and use self-attention to learn image-text semantic alignments in a brute force manner, in this paper, we propose a new learning method Oscar (Object-Semantics Aligned Pre-training), which uses object tags detected in images as anchor points to significantly ease the learning of alignments.
The paper shared the most on social media this week is by a team at University of Tübingen: "Shortcut Learning in Deep Neural Networks" by Robert Geirhos et al (Apr 2020)
The most influential Twitter user discussing papers is Horace Dediu who shared "Generalized logistic growth modeling of the COVID-19 outbreak in 29 provinces in China and in the rest of the world" by Ke Wu et al (Mar 2020)
This week was active for "Computer Science - Computers and Society", with 30 new papers.
The paper discussed most in the news over the past week was by a team at Boston University: "Anonymous Collocation Discovery: Harnessing Privacy to Tame the Coronavirus" by Ran Canetti et al (Mar 2020), which was referenced 10 times, including in the article How Reliable and Effective Are the Mobile Apps Being Used to Fight COVID-19? in Wire. The paper author, Ari Trachtenberg (Boston University), was quoted saying "When a person is tested positive for COVID-19, the person could choose (through the administrating medical professional) to voluntarily share their list of random numbers -- either their own generated numbers or the numbers that the app observed". The paper got social media traction with 57 shares. On Twitter, @alexcryptan said "By now there are at least 5 academic initiatives: ● TCN ● Canetti et al. ● DP-3T (European) ● PACT ● MIT protocol".
Leading researcher Yoshua Bengio (Université de Montréal) came out with "Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims", which had 60 shares over the past 3 days. @ed_teather tweeted "This paper on trustworthy & accountable AI is excellent on how to move ethics from principles to practice. Particularly like the idea of putting a bounty on bias & developing 3rd party audits. Congrats to all those involved 👏👏". This paper was also shared the most on social media with 305 tweets. @ed_teather (Ed Teather) tweeted "This paper on trustworthy & accountable AI is excellent on how to move ethics from principles to practice. Particularly like the idea of putting a bounty on bias & developing 3rd party audits. Congrats to all those involved 👏👏".
Over the past week, 17 new papers were published in "Computer Science - Human-Computer Interaction".
The paper discussed most in the news over the past week was "iCub: Learning Emotion Expressions using Human Reward" by Nikhil Churamani et al (Mar 2020), which was referenced 1 time, including in the article Teaching the iCub robot to express basic human emotions in Tech Xplore. The paper got social media traction with 6 shares. A Twitter user, @NikhilChuramani, commented "Old, but still an update. Extended abstract about Learning Emotion Expressions using Human Reward on the iCub robot from IROS 2016 Workshop now available on Joint work with #FranciscoCruz and".
The paper shared the most on social media this week is by a team at Adobe: "Intuitive, Interactive Beard and Hair Synthesis with Generative Models" by Kyle Olszewski et al (Apr 2020) with 51 shares.
This week was active for "Computer Science - Learning", with 256 new papers.
The paper discussed most in the news over the past week was by a team at Google: "TensorFlow Quantum: A Software Framework for Quantum Machine Learning" by Michael Broughton et al (Mar 2020), which was referenced 33 times, including in the article Best of arXiv.org for AI, Machine Learning, and Deep Learning – March 2020 in InsideBIGDATA. The paper author, Michael Broughton (University of Waterloo), was quoted saying "there is no strong indication that we should expect a quantum advantage for the classification of classical data using QNNs in the near term." The paper got social media traction with 299 shares. A user, @quantumVerd, tweeted "For those technically inclined, here is the Quantum paper - which includes technical details on both the software architecture and underlying theory. 👉 Included is an extensive section full of #TFQuantum example notebooks".
Leading researcher Sergey Levine (University of California, Berkeley) published "Datasets for Data-Driven Reinforcement Learning" @faoliehoek tweeted "That's a strange statement... I think it's actually *true online* RL that we need for the real world...!".
The paper shared the most on social media this week is by a team at MIT: "At the Interface of Algebra and Statistics" by Tai-Danae Bradley (Apr 2020) with 627 shares. @seanjtaylor (Sean J. Taylor) tweeted "I wish I were a fraction as good at explaining complex ideas as Tai-Danae. She has an amazing gift, I'm looking forward to checking out her thesis. This blog post from last year transformed how I think about matrices: (don't miss the sequel to that post!)".
The most influential Twitter user discussing papers is Horace Dediu who shared "Generalized logistic growth modeling of the COVID-19 outbreak in 29 provinces in China and in the rest of the world" by Ke Wu et al (Mar 2020)
Over the past week, nine new papers were published in "Computer Science - Multiagent Systems".
Over the past week, 23 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 Google: "AutoML-Zero: Evolving Machine Learning Algorithms From Scratch" by Esteban Real et al (Mar 2020)
Over the past week, 37 new papers were published in "Computer Science - Robotics".
The paper discussed most in the news over the past week was by a team at Carnegie Mellon University: "Learning to Explore using Active Neural SLAM" by Devendra Singh Chaplot et al (Apr 2020), which was referenced 4 times, including in the article Facebook’s AI teaches robots to navigate environments using less data in Venturebeat. The paper also got the most social media traction with 252 shares. On Twitter, @wgussml observed "Amazing to see how much progress has made on the hard problem of active SLAM in the past couple of years!", while @popular_ML observed "The most popular ArXiv tweet in the last 24h".
Leading researcher Sergey Levine (University of California, Berkeley) came out with "Thinking While Moving: Deep Reinforcement Learning with Concurrent Control"