Week Ending 3.22.2020
RESEARCH WATCH: 3.22.2020
Over the past week, 65 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 New York University: "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" by Gary Marcus (Feb 2020), which was referenced 7 times, including in the article Hybrid AI systems are quietly solving the problems of deep learning in The Next Web. The paper author, Gary Marcus (New York University), was quoted saying "There are plenty of first steps towards building architectures that combine the strengths of the symbolic approaches with insights from machine learning, in order to develop better techniques for extracting and generalizing abstract knowledge from large, often noisy data sets". The paper got social media traction with 1105 shares. A Twitter user, @AgileThought, posted "What can we expect from during the next decade? Hint: Even more reliable, robust AI. Learn more".
Leading researcher Sergey Levine (University of California, Berkeley) came out with "Sequential Forecasting of 100,000 Points" The authors study the problem of future prediction at the level of 3D scenes, represented by point clouds captured by a LiDAR sensor, i.e., directly learning to forecast the evolution of >100,000 points that comprise a complete scene.
The paper shared the most on social media this week is by a team at University of Science and Technology of China: "A Survey on Contextual Embeddings" by Qi Liu et al (Mar 2020) with 68 shares. The researchers review existing contextual embedding models, cross - lingual polyglot pre - training, the application of contextual embeddings in downstream tasks, model compression, and model analyses. @omarsar0 (elvis) tweeted "This week saw a couple of nice papers to help you get up to speed with contextualized embeddings and language models. Please share if you come across others. paper 1: paper 2: paper 3".
This week was very active for "Computer Science - Computer Vision and Pattern Recognition", with 323 new papers.
The paper discussed most in the news over the past week was by a team at Microsoft: "Inverse Graphics GAN: Learning to Generate 3D Shapes from Unstructured 2D Data" by Sebastian Lunz et al (Feb 2020), which was referenced 6 times, including in the article This clever new algorithm can make 3D objects from 2D photos in RedShark News. The paper got social media traction with 32 shares. A Twitter user, @ankurt, observed "This fascinating latest work from is the tip of the spear for 2D to 3D rendering. If one could scale this up, the applications are limitless. Kudos!", while @jbohnslav commented "Inverse graphics GAN: learning to generate 3D shapes from unstructured 2D data Pretty interesting use of OpenGL + a differentiable approximation module, instead of differentiable renderers!".
Leading researcher Yoshua Bengio (Université de Montréal) published "Object-Centric Image Generation from Layouts" The investigators start with the idea that a model must be able to understand individual objects and relationships between objects in order to generate complex scenes well.
The paper shared the most on social media this week is by a team at UC Berkeley: "NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis" by Ben Mildenhall et al (Mar 2020) with 1420 shares. @jongranskog (Jonathan Granskog) tweeted "I like the ray marching approach with volumetric rendering. But it has to be trained for at least 12 hours per scene and takes 30s to evaluate".
Over the past week, 27 new papers were published in "Computer Science - Computers and Society".
Over the past week, 23 new papers were published in "Computer Science - Human-Computer Interaction".
The paper discussed most in the news over the past week was by a team at IBM: "Do ML Experts Discuss Explainability for AI Systems? A discussion case in the industry for a domain-specific solution" by Juliana Jansen Ferreira et al (Feb 2020), which was referenced 1 time, including in the article IBM at the Intersection of Human-Computer Interaction and AI in IBM Research. The paper got social media traction with 6 shares.
This week was very active for "Computer Science - Learning", with 282 new papers.
The paper discussed most in the news over the past week was by a team at New York University: "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence" by Gary Marcus (Feb 2020)
Leading researcher Yoshua Bengio (Université de Montréal) published "Object-Centric Image Generation from Layouts"
The paper shared the most on social media this week is "Neuroevolution of Self-Interpretable Agents" by Yujin Tang et al (Mar 2020) with 432 shares. @hardmaru (hardmaru) tweeted "Neuroevolution of Self-Interpretable Agents Agents with a self-attention “bottleneck” not only can solve these tasks from pixel inputs with only 4000 parameters, but they are also better at generalization! article pdf Read on 👇🏼".
This week was active for "Computer Science - Multiagent Systems", with 24 new papers.
The paper discussed most in the news over the past week was "Ford Multi-AV Seasonal Dataset" by Siddharth Agarwal (Ford AV LLC) et al (Mar 2020), which was referenced 3 times, including in the article Ford releases a data set to accelerate autonomous car development in Venturebeat. The paper also got the most social media traction with 51 shares. The researchers present a challenging multi - agent seasonal dataset collected by a fleet of Ford autonomous vehicles at different days and times during 2017 - 18. A user, @OpenRoboticsOrg, tweeted "Did you catch the huge all-weather autonomous vehicle ROSBag data set that released yesterday? This morning they released the pre-print publication. You should check it out!".
Leading researcher Sergey Levine (University of California, Berkeley) published "Sequential Forecasting of 100,000 Points" The researchers study the problem of future prediction at the level of 3D scenes, represented by point clouds captured by a LiDAR sensor, i.e., directly learning to forecast the evolution of >100,000 points that comprise a complete scene.
Over the past week, 24 new papers were published in "Computer Science - Neural and Evolutionary Computing".
The paper discussed most in the news over the past week was "Evolved Neuromorphic Control for High Speed Divergence-based Landings of MAVs" by J. J. Hagenaars et al (Mar 2020), which was referenced 1 time, including in the article Neuromorphic controllers to enhance control during landing in micro air vehicles in Tech Xplore. The paper author, Jesse Hagenaars, was quoted saying "The final goal of our research is to connect motion estimation and control to end up with a completely bio-inspired pipeline, which will be much more efficient in terms of energy spending than traditional vision-based control approaches". The paper was shared 2 times in social media.
Leading researcher J. Warner (University of Florida) came out with "Using context to make gas classifiers robust to sensor drift" The researchers focus on extending the lifetime of sensor systems without additional data acquisition by transfering knowledge from one time window to a subsequent one after drift has occurred.
The paper shared the most on social media this week is "Neuroevolution of Self-Interpretable Agents" by Yujin Tang et al (Mar 2020)
This week was active for "Computer Science - Robotics", with 58 new papers.
The paper discussed most in the news over the past week was "Ford Multi-AV Seasonal Dataset" by Siddharth Agarwal (Ford AV LLC) et al (Mar 2020)
Leading researcher Sergey Levine (University of California, Berkeley) published "Sequential Forecasting of 100,000 Points"