Week Ending 5.2.2021
RESEARCH WATCH: 5.2.2021
This week was active for "Computer Science", with 1,186 new papers.
The paper discussed most in the news over the past week was "Detecting Hate Speech with GPT-3" by Ke-Li Chiu et al (Mar 2021), which was referenced 28 times, including in the article 'Can I see your parts list?' What AI's attempted chat-up lines tell us about computer-generated language in Yahoo! News UK and Ireland. The paper got social media traction with 21 shares.
Leading researcher Ruslan Salakhutdinov (Carnegie Mellon University) came out with "A Note on Connecting Barlow Twins with Negative-Sample-Free Contrastive Learning".
The paper shared the most on social media this week is "MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding" by Aishwarya Kamath et al (Apr 2021) with 52 shares. The investigators propose MDETR, an end - to - end modulated detector that detects objects in an image conditioned on a raw text query, like a caption or a question. @giffmana (Lucas Beyer) tweeted "The pink elephant (not a mask, actual pixels) is a really neat demo idea, I love it! Whoever though of it deserves a nice bonus ;-)".
This week was very active for "Computer Science - Artificial Intelligence", with 162 new papers.
The paper discussed most in the news over the past week was "Measuring Mathematical Problem Solving With the MATH Dataset" by Dan Hendrycks et al (Mar 2021), which was referenced 16 times, including in the article Huawei trained the Chinese-language equivalent of GPT-3 in Venturebeat. The paper got social media traction with 195 shares. A Twitter user, @DanHendrycks, observed "To find the limits of Transformers, we collected 12,500 math problems. While a three-time IMO gold medalist got 90%, GPT-3 models got ~5%, with accuracy increasing slowly. If trends continue, ML models are far from achieving mathematical reasoning".
This week was very active for "Computer Science - Computer Vision and Pattern Recognition", with 311 new papers.
The paper discussed most in the news over the past week was by a team at MIT: "The ThreeDWorld Transport Challenge: A Visually Guided Task-and-Motion Planning Benchmark for Physically Realistic Embodied AI" by Chuang Gan et al (Mar 2021), which was referenced 16 times, including in the article Reinforcement learning competition pushes the boundaries of embodied AI in Venturebeat. The paper author, Chuang Gan (IBM), was quoted saying "This environment can be used to train RL models, which fall short on these types of tasks and require explicit reasoning and planning abilities". The paper got social media traction with 23 shares. On Twitter, @gan_chuang said "Excited to announce the TDW Transport Challenge in the CVPR21 Embodied AI Workshop! Can you train agents to change object states and fulfill complex goals in simulated physical environments? Paper Link: Stater Code".
Leading researcher Yann LeCun (New York University) came out with "MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding" The authors propose MDETR, an end - to - end modulated detector that detects objects in an image conditioned on a raw text query, like a caption or a question. @giffmana tweeted "The pink elephant (not a mask, actual pixels) is a really neat demo idea, I love it! Whoever though of it deserves a nice bonus ;-)".
The paper shared the most on social media this week is by a team at University of Chinese Academy of Sciences: "EigenGAN: Layer-Wise Eigen-Learning for GANs" by Zhenliang He et al (Apr 2021) with 52 shares.
Over the past week, 29 new papers were published in "Computer Science - Computers and Society".
The paper discussed most in the news over the past week was by a team at Google: "Carbon Emissions and Large Neural Network Training" by David Patterson et al (Apr 2021), which was referenced 17 times, including in the article POLITICO Pro AI: Decoded: Washington’s AI awakening — Europe split on facial recognition — What to do about health data in Politico.eu. The paper author, David Patterson (University of California, Santa Barbara), was quoted saying "is equivalent to roughly 200,000 to 300,000 whole passenger jet SF↔NY round trips". The paper got social media traction with 610 shares. On Twitter, @timnitGebru posted "This is like a never ending nightmare. Its just so unreal. A paper written by ~90% men authors (8 out 9 authors are men), one of whom fired co-authors from 100% underrepresented groups, for writing about environmental racism, sexism and other issues that affect us".
This week was active for "Computer Science - Human-Computer Interaction", with 28 new papers.
The paper discussed most in the news over the past week was "Design and Test of an adaptive augmented reality interface to manage systems to assist critical missions" by Dany Naser Addin et al (Mar 2021), which was referenced 5 times, including in the article New Headsets Let Police Control Drone Swarms 'Hands Free' in Interesting Engineering. The paper author, Dany Naser Addin, was quoted saying "The technology we developed can bring a huge flow of information that can overload the user and must thus be filtered in an optimal way, in order to improve the situational awareness of the user and help him/her to understand the current situation effectively". The paper was shared 2 times in social media.
This week was very active for "Computer Science - Learning", with 368 new papers.
The paper discussed most in the news over the past week was by a team at Google: "Carbon Emissions and Large Neural Network Training" by David Patterson et al (Apr 2021)
Leading researcher Ruslan Salakhutdinov (Carnegie Mellon University) published "A Note on Connecting Barlow Twins with Negative-Sample-Free Contrastive Learning".
The paper shared the most on social media this week is "MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding" by Aishwarya Kamath et al (Apr 2021)
Over the past week, 12 new papers were published in "Computer Science - Multiagent Systems".
Over the past week, 21 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 IBM: "Learning in Deep Neural Networks Using a Biologically Inspired Optimizer" by Giorgia Dellaferrera et al (Apr 2021), which was referenced 2 times, including in the article Toward a New Generation of Neuromorphic Computing: IBM & ETH Zurich’s Biologically Inspired Optimizer Boosts FCNN and SNN Training in SyncedReview.com. The paper got social media traction with 5 shares. The authors propose a novel biologically inspired optimizer for artificial (ANNs) and spiking neural networks (SNNs) that incorporates key principles of synaptic integration observed in dendrites of cortical neurons : GRAPES (Group Responsibility for Adjusting the Propagation of Error Signals).
This week was active for "Computer Science - Robotics", with 63 new papers.
The paper discussed most in the news over the past week was by a team at MIT: "The ThreeDWorld Transport Challenge: A Visually Guided Task-and-Motion Planning Benchmark for Physically Realistic Embodied AI" by Chuang Gan et al (Mar 2021)