Week Ending 8.30.2020

 

RESEARCH WATCH: 8.30.2020

 
ai-research.png

This week was active for "Computer Science - Artificial Intelligence", with 105 new papers.

This week was active for "Computer Science - Computer Vision and Pattern Recognition", with 246 new papers.

  • The paper discussed most in the news over the past week was "Rewriting a Deep Generative Model" by David Bau et al (Jul 2020), which was referenced 7 times, including in the article What should the world look like? Some neural nets have their own answer, find researchers in ZDNet. The paper author, Antonio Torralba (Massachusetts Institute of Technology), was quoted saying "Even this relationship suggests that associations learned from data can be stored as lines of memory, and not only located but reversed". The paper got social media traction with 116 shares. The investigators introduce a new problem setting : manipulation of specific rules encoded by a deep generative model. A user, @skybase, tweeted "Hold up hold up lol... please lol. whaat", while @atsushieeeee posted "That's an amazing research similar to styleGAN2 distillation! Some examples are so practical and others are humorous!".

  • Leading researcher Yoshua Bengio (Université de Montréal) published "Visual Concept Reasoning Networks".

  • The paper shared the most on social media this week is "Anime-to-Real Clothing: Cosplay Costume Generation via Image-to-Image Translation" by Koya Tango et al (Aug 2020) with 133 shares. @noop_noob (Amateur Slacker) tweeted "So someone tried to make computers extract the clothing from anime images for... cosplaying purposes. Why tho? (In the images, top row is input, bottom row is computer-generated output, middle row is actual photo for reference.)".

This week was active for "Computer Science - Computers and Society", with 38 new papers.

  • The paper discussed most in the news over the past week was by a team at Stanford University: "Analyzing Who and What Appears in a Decade of US Cable TV News" by James Hong et al (Aug 2020), which was referenced 5 times, including in the article Interactive tool uses AI to search transcripts and calculate the screen time of public figures in Tech Xplore. The paper author, Maneesh Agrawala (Stanford University), was quoted saying "By letting researchers, journalists and the public quantitatively measure who and what is in the news, the tool can help identify biases and trends in cable TV news coverage". The paper got social media traction with 10 shares. The authors use computational techniques to analyze a data set of nearly 24/7 video, audio, and text captions from three major U.S. cable TV networks (CNN, FOX News, and MSNBC) from the last decade. A Twitter user, @MattGrossmann, said "Trump has dominated cable news coverage unlike any other figure. Evidence from automated screen capture. Women have also increased to a majority over time. #SocSciResearch".

  • The paper shared the most on social media this week is "The Lockdown Effect: Implications of the COVID-19 Pandemic on Internet Traffic" by Anja Feldmann et al (Aug 2020) with 68 shares. The investigators In this paper, using data from a diverse set of vantage points (one ISP, three IXPs, and one metropolitan educational network), we study the effect of these lockdowns on traffic shifts. @jtkristoff (jtk) tweeted "Weekend Reads: * - side channel attacks * - Internet traffic and COVID-19 * - GitHub Ruby upgrade * - clean slate web * - making IETF RFCs".

This week was active for "Computer Science - Human-Computer Interaction", with 35 new papers.

This week was very active for "Computer Science - Learning", with 328 new papers.

  • The paper discussed most in the news over the past week was "Rewriting a Deep Generative Model" by David Bau et al (Jul 2020)

  • Leading researcher Yoshua Bengio (Université de Montréal) came out with "Visual Concept Reasoning Networks".

  • The paper shared the most on social media this week is by a team at Massachusetts Institute of Technology: "The Hessian Penalty: A Weak Prior for Unsupervised Disentanglement" by William Peebles et al (Aug 2020) with 124 shares. The investigators propose the Hessian Penalty, a simple regularization term that encourages the Hessian of a generative model with respect to its input to be diagonal. @xsteenbrugge (Xander Steenbrugge) tweeted "Super cool! What you're doing is also related to this approach: I feel like the dark & murky latent mechanics of Deep Generative models are becoming clearer by the day! I'm currently implementing this approach for StyleGANv2".

Over the past week, 11 new papers were published in "Computer Science - Multiagent Systems".

Over the past week, 12 new papers were published in "Computer Science - Neural and Evolutionary Computing".

This week was active for "Computer Science - Robotics", with 59 new papers.


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