Week Ending 12.12.2021
RESEARCH WATCH: 12.12.2021
This week was extremely active for "Computer Science - Artificial Intelligence", with 310 new papers.
The paper discussed most in the news over the past week was by a team at Massachusetts Institute of Technology: "3DP3: 3D Scene Perception via Probabilistic Programming" by Nishad Gothoskar et al (Oct 2021), which was referenced 11 times, including in the article Machines that see world more like humans do in Mirage News. The paper author, Nishad Gothoskar, was quoted saying "If you don’t know about the contact relationships, then you could say that an object is floating above the table – that would be a valid explanation. As humans, it is obvious to us that this is physically unrealistic and the object resting on top of the table is a more likely pose of the object. Because our reasoning system is aware of this sort of knowledge, it can infer more accurate poses. That is a key insight of this work". The paper was shared 3 times in social media. A Twitter user, @summarizedml, commented "3DP3, a framework for inverse graphics that uses inference in a structured generative model of objects, scenes, and images. 📄".
Leading researcher Yoshua Bengio (Université de Montréal) published "Multi-scale Feature Learning Dynamics: Insights for Double Descent" The investigators investigate the origins of the less studied epoch - wise double descent in which the test error undergoes two non - monotonous transitions, or descents as the training time increases. @summarizedml tweeted "We investigate the origins of epoch-wise double descent in which the test error undergoes two non-monotonous transitions, or descents 📄".
The paper shared the most on social media this week is by a team at DeepMind: "Player of Games" by Martin Schmid et al (Dec 2021) with 388 shares. @PatrickPilarski (Patrick M. Pilarski) tweeted "Excellent new research from our Edmonton DeepMind office, showing an agent that can learn to skillfully engage in perfect and imperfect information games, including Scotland Yard. Great work folks!".
The most influential Twitter user discussing papers is Francis Villatoro who shared "Mercury as the relic of Earth and Venus outward migration" by Matthew S. Clement et al (Nov 2021) and said: "Mercury as the relic of Earth and Venus' outward migration the migration of giant planet cores may have reshaped the surface density profile of terrestrial planet-forming material and generated conditions for the formation of Mercury-like planets".
This week was very active for "Computer Science - Computer Vision and Pattern Recognition", with 373 new papers.
The paper discussed most in the news over the past week was by a team at Massachusetts Institute of Technology: "3DP3: 3D Scene Perception via Probabilistic Programming" by Nishad Gothoskar et al (Oct 2021)
Leading researcher Sergey Levine (University of California, Berkeley) came out with "Extending the WILDS Benchmark for Unsupervised Adaptation" The researchers present the WILDS 2.0 update, which extends 8 of the 10 datasets in the WILDS benchmark of distribution shifts to include curated unlabeled data that would be realistically obtainable in deployment. @ak92501 tweeted "Extending the WILDS Benchmark for Unsupervised Adaptation abs: Wilds 2.0 update, extends 8 of the 10 datasets in the Wilds benchmark of distribution shifts to include curated unlabeled data that would be realistically obtainable in deployment".
The paper shared the most on social media this week is by a team at UC Berkeley: "Plenoxels: Radiance Fields without Neural Networks" by Alex Yu et al (Dec 2021) with 416 shares. @dsilvavinicius (Vinícius da Silva) tweeted "Incredible. Now we can train NeRFs in 10 or 20 minutes".
Over the past week, 22 new papers were published in "Computer Science - Computers and Society".
The paper discussed most in the news over the past week was "Bugs in our Pockets: The Risks of Client-Side Scanning" by Hal Abelson et al (Oct 2021), which was referenced 32 times, including in the article Germany’s new government will firmly defend encryption, key Social Democrat says in EurActiv.com. The paper author, Troncoso, was quoted saying "The checks and balances that limit the scope of previous surveillance methods in democracies just aren’t there with broad deployment of CSS. As law-abiding citizens, we should be free to use our devices to make our lives easier, without worry of being bugged like a spy movie villain". The paper also got the most social media traction with 905 shares. On Twitter, @AlecMuffett posted "1/ THE MOST IMPORTANT information security discussion of the day will be about the publication of #BugsInOurPockets, a paper by the biggest names in encryption, regarding #ClientSideScanning and the likes of the #CSAM-detection proposal".
This week was very active for "Computer Science - Human-Computer Interaction", with 46 new papers.
This week was extremely active for "Computer Science - Learning", with 459 new papers.
The paper discussed most in the news over the past week was by a team at Massachusetts Institute of Technology: "ExoMiner: A Highly Accurate and Explainable Deep Learning Classifier that Validates 301 New Exoplanets" by Hamed Valizadegan et al (Nov 2021), which was referenced 82 times, including in the article NASA announces the discovery of 301 new exoplanets in Archynewsy. The paper author, Hamed Valizadegan (Machine learning manager with the Universities Space Research Association at Ames), was quoted saying "When ExoMiner says something is a planet, you can be sure it’s a planet. ExoMiner is highly accurate and in some ways more reliable than both existing machine classifiers and the human experts it’s meant to emulate because of the biases that come with human labeling. Now that we’ve trained ExoMiner using Kepler data, with a little fine-tuning, we can transfer that learning to other missions, including TESS, which we’re currently working on. There’s room to grow." The paper got social media traction with 41 shares. A Twitter user, @storybywill, commented "We didn't. The team who developed ExoMine did, and here is their paper", while @arxiv_cs_LG observed "ExoMiner: A Highly Accurate and Explainable Deep Learning Classifier to Mine Exoplanets. Valizadegan, Martinho, Wilkens, Jenkins, Smith, Caldwell, Twicken, Gerum, Walia, Hausknecht, Lubin, Bryson, and Oza".
Leading researcher Yoshua Bengio (Université de Montréal) came out with "Multi-scale Feature Learning Dynamics: Insights for Double Descent" The researchers investigate the origins of the less studied epoch - wise double descent in which the test error undergoes two non - monotonous transitions, or descents as the training time increases. @summarizedml tweeted "We investigate the origins of epoch-wise double descent in which the test error undergoes two non-monotonous transitions, or descents 📄".
The paper shared the most on social media this week is by a team at DeepMind: "Player of Games" by Martin Schmid et al (Dec 2021)
The most influential Twitter user discussing papers is Francis Villatoro who shared "Mercury as the relic of Earth and Venus outward migration" by Matthew S. Clement et al (Nov 2021)
Over the past week, 13 new papers were published in "Computer Science - Multiagent Systems".
The paper discussed most in the news over the past week was by a team at Tianjin University: "SEIHAI: A Sample-efficient Hierarchical AI for the MineRL Competition" by Hangyu Mao et al (Nov 2021), which was referenced 2 times, including in the article SEIHAI: The hierarchical AI that won the NeurIPS-2020 MineRL competition in Tech Xplore. The paper author, Hangyu Mao et al, was quoted saying "We present SEIHAI, a sample-efficient hierarchical AI that fully takes advantage of the human demonstrations and the task structure". The paper got social media traction with 8 shares. A user, @summarizedml, tweeted "A new algorithm that can efficiently leverage humandemonstrations to solve the MineRL task with sparse rewards. 📄", while @gastronomy posted "> The MineRL competition is designed for the development of reinforcement learning and imitation learning algorithms that can efficiently lever".
Over the past week, 18 new papers were published in "Computer Science - Neural and Evolutionary Computing".
The paper discussed most in the news over the past week was "Recurrent neural network models for working memory of continuous variables: activity manifolds, connectivity patterns, and dynamic codes" by Christopher J. Cueva et al (Nov 2021), which was referenced 1 time, including in the article Stay updated with Neuroscience: November 2021 Must-Reads in Towards Data Science. The paper got social media traction with 15 shares.
This week was very active for "Computer Science - Robotics", with 70 new papers.
The paper discussed most in the news over the past week was by a team at Tianjin University: "SEIHAI: A Sample-efficient Hierarchical AI for the MineRL Competition" by Hangyu Mao et al (Nov 2021)
Leading researcher Sergey Levine (University of California, Berkeley) published "CoMPS: Continual Meta Policy Search" @HochreiterSepp tweeted "ArXiv Sequential multi-task learning without revisiting previous tasks. Uses memory of successful trajectories from older tasks for behavioral cloning. We also memorized successful episodes in Hopfield-RUDDER (NeurIPS RL workshops)".