Eye On AI

View Original

Week Ending 03.31.19

RESEARCH WATCH: 03.31.19

Over the past week, 205 new papers were published in "Computer Science".

  • The paper discussed most in the news over the past week was by a team at Ben-Gurion University of the Negev: "SoK - Security and Privacy in the Age of Drones: Threats, Challenges, Solution Mechanisms, and Scientific Gaps" by Ben Nassi et al (Mar 2019), which was referenced 9 times, including in the article As drones fill the skies, cybercriminals won’t be far behind in ProfessionalHackers.in. The paper author, Ben Nassi (Ph.D. student in BGU ’s  Department of Software and Information Systems Engineering ( SISE ) and a researcher at the BGU Cyber Security Research Center), was quoted saying "The cutting-edge technology and decreasing drone prices made them accessible to individuals and organizations, but has created new threats and recently caused an increase in drone-related incidents". The paper was shared 3 times in social media.

  • Leading researcher Jianfeng Gao (Microsoft) came out with "Cyclical Annealing Schedule: A Simple Approach to Mitigating KL Vanishing" @gastronomy tweeted "> Variational autoencoders (VAEs) with an auto-regressive decoder have been applied for many natural language processing (NLP) tasks. The V".

  • The paper shared the most on social media this week is by a team at Johns Hopkins University: "Weight Standardization" by Siyuan Qiao et al (Mar 2019) with 151 shares. The investigators propose Weight Standardization (WS) to accelerate deep network training. @mosko_mule (ワクワクさん) tweeted "Weight Standardization BNはバッチサイズが小さきには有効ではないがGNなどの他のactivationの正規化はBNほど有効ではなかった。重みを正規化するWSはBatch Size=1などの極端な状況でも有効でGNと組み合わせることで大BSでのBNを様々なタスクで凌駕。Simple is Best".

Over the past week, 72 new papers were published in "Computer Science - Artificial Intelligence".

This week was active for "Computer Science - Computer Vision and Pattern Recognition", with 213 new papers.

Over the past week, 15 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 Ben-Gurion University of the Negev: "SoK - Security and Privacy in the Age of Drones: Threats, Challenges, Solution Mechanisms, and Scientific Gaps" by Ben Nassi et al (Mar 2019), which was referenced 9 times, including in the article As drones fill the skies, cybercriminals won’t be far behind in ProfessionalHackers.in. The paper author, Ben Nassi (Ph.D. student in BGU ’s  Department of Software and Information Systems Engineering ( SISE ) and a researcher at the BGU Cyber Security Research Center), was quoted saying "The cutting-edge technology and decreasing drone prices made them accessible to individuals and organizations, but has created new threats and recently caused an increase in drone-related incidents". The paper was shared 3 times in social media.

  • The paper shared the most on social media this week is by a team at University College London: "Airbnbs disruption of the housing structure in London" by Zahratu Shabrina et al (Mar 2019) with 89 shares. The researchers explore Airbnb, a peer - to - peer platform for short - term rental of housing accommodation, examining the geographical pattern of those establishments using data from London. @bwaber (Ben Waber) tweeted "disruption of the housing structure in London Fascinating analysis out of Key quote: ... detracts more than 1.4% of the housing supply into short-term rentals. Such a phenomenon can reach up to 20% in some neighbourhoods !!!".

Over the past week, 13 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 University of California, Los Angeles: "VRKitchen: an Interactive 3D Virtual Environment for Task-oriented Learning" by Xiaofeng Gao et al (Mar 2019), which was referenced 1 time, including in the article VRKitchen: An interactive virtual environment to train and test AI agents in PhysOrg.com. The paper author, Tianmin Shu (University of California, Los Angeles), was quoted saying "Popular approaches in recent literature that work well in simpler virtual environments fail to achieve decent performance on the cooking tasks implemented in our platform, imposing a new challenge for AI research and highlighting the importance of building more realistic and complex virtual environments". The paper got social media traction with 27 shares. A user, @yapp1e, tweeted "VRKitchen: an Interactive 3D Virtual Environment for Task-oriented Learning. One of the main challenges of advancing task-oriented learning such as visual task planning and reinforcement learning is the lack of realistic and", while @gastronomy observed "> One of the main challenges of advancing task-oriented learning such as visual task planning and reinforcement learning is the lack of r".

Over the past week, 149 new papers were published in "Computer Science - Learning".

  • The paper discussed most in the news over the past week was by a team at Google: "Model-Based Reinforcement Learning for Atari" by Lukasz Kaiser et al (Mar 2019), which was referenced 27 times, including in the article Google researchers improve reinforcement learning by having their AI play Pong in Venturebeat. The paper author, Henryk Michalewski (University of Warsaw), was quoted saying "This is one of the important ideas of reinforcement learning. A recent survey done by DeepMind's J.B. Hammrick provides a thorough account of analogies between model-based reinforcement learning and mental simulation as considered by cognitive science". The paper got social media traction with 380 shares. A Twitter user, @InsolentAI, observed "Latest #ML #reinforcementlearning research: agent becomes superhuman #gamer 🎮🕹 on #Atari in only two hours of real-time gameplay!", while @arankomatsuzaki observed "Model-Based Reinforcement Learning for Atari: Achieving human-level performance on many Atari games after two hours of real-time play".

  • Leading researcher Yoshua Bengio (Université de Montréal) came out with "Wasserstein Dependency Measure for Representation Learning" @hillbig tweeted "Although energy-based mutual information lower-bound is used for representation learning, this estimation requires prohibitive sample size when MI is high. They propose to use the Wasserstein distance instead of KL and show it's effectiveness".

  • The paper shared the most on social media this week is "Yet Another Accelerated SGD: ResNet-50 Training on ImageNet in 74.7 seconds" by Masafumi Yamazaki et al (Mar 2019) with 178 shares. @ankurhandos (Ankur Handa) tweeted "Just another day at the office for some at Fujitsu Laboratories. The original AlexNet took 1 week to train on ImageNet on a single GPU and now it takes ~1.2 minutes to train ImageNet. The training throughput is over 1.73 million images/sec 🤯".

Over the past week, five new papers were published in "Computer Science - Multiagent Systems".

Over the past week, 29 new papers were published in "Computer Science - Neural and Evolutionary Computing".

This week was active for "Computer Science - Robotics", with 50 new papers.





EYE ON A.I. GETS READERS UP TO DATE ON THE LATEST FUNDING NEWS AND RELATED ISSUES. SUBSCRIBE FOR THE WEEKLY NEWSLETTER.