Week Ending 08.25.19
RESEARCH WATCH: 08.25.19
Over the past week, 1,001 new papers were published in "Computer Science".
The paper discussed most in the news over the past week was "Artificial Intelligence and the Future of Psychiatry: Insights from a Global Physician Survey" by P. Murali Doraiswamy et al (Jul 2019), which was referenced 41 times, including in the article Are psychiatrists really ready for the AI revolution? in Technology Review. The paper author, Murali Doraiswamy, was quoted saying "It is time for us to stop thinking about AI as a battle of machines versus humans. We need to instead focus on how AI can optimize and improve clinicians’ abilities to deliver better care".
Leading researcher Kyunghyun Cho (New York University) came out with "Dynamics-aware Embeddings", which has 0 shares on Twitter so far. The researchers consider self - supervised representation learning to improve sample efficiency in reinforcement learning (RL).
Over the past week, 70 new papers were published in "Computer Science - Artificial Intelligence".
The paper discussed most in the news over the past week was "Playing magic tricks to deep neural networks untangles human deception" by Regina Zaghi-Lara et al (Aug 2019), which was referenced 2 times, including in the article Researchers attempt to fool AI with magic tricks in Venturebeat. The paper got social media traction with 58 shares. The investigators deconstructed stage magic into purely motor maneuvers and trained an artificial neural network (DeepLabCut) to follow coins as a professional magician made them appear and disappear in a series of tricks. On Twitter, @TrackingActions commented "A clever idea from to see when an model is tricked vs human(s) tricked. 🎩🤹🏻♂️🔮#deeplabcut #deeplabMAGIC Check out the cool video", while @ThomasFraps commented "AI can be fooled, e.g. but only knowledge of the world gives rise to the illusion of impossibility: „It is a world of difference between a spectator’s not knowing how something is done and knowing that it can‘t be done.“ - Simon Aronson #magicturingtestAGI".
Leading researcher Kyunghyun Cho (New York University) came out with "Dynamics-aware Embeddings". This paper was also shared the most on social media with 84 tweets. @BayesForDays (Cassandra Jacobs, Subtweeter in Chief) tweeted "Y'all this compositionality in neural networks paper is so good".
This week was active for "Computer Science - Computer Vision and Pattern Recognition", with 271 new papers.
The paper discussed most in the news over the past week was "U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation" by Junho Kim et al (Jul 2019), which was referenced 9 times, including in the article This AI Can Turn Your Selfies into Anime Characters in Beebom. The paper author, Minjae Kim (Université Paris-Saclay), was quoted saying "I was inspired by a paper called ‘class activation mapping’ (CAM)". The paper got social media traction with 357 shares. On Twitter, @tarrysingh commented "We live in an age of discoveries and fun with #AI This tool can now change your photo to #anime character. I’ll surely try it on my photo soon and post it here 😊 Code: Paper: For continuous updates on #A".
Leading researcher Dhruv Batra (Georgia Institute of Technology) came out with "Sequential Latent Spaces for Modeling the Intention During Diverse Image Captioning".
The paper shared the most on social media this week is "Generating High-Resolution Fashion Model Images Wearing Custom Outfits" by Gökhan Yildirim et al (Aug 2019) with 129 shares. @h_schreiber (Hendrik Schreiber) tweeted "Crushing the dreams of millions of little princesses. Who still needs fashion models when you have GANs?!".
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 University of Copenhagen: "Tracking Behavioral Patterns among Students in an Online Educational System" by Stephan Lorenzen et al (Aug 2019), which was referenced 2 times, including in the article Researchers use AI to track students' performance in online courses in Venturebeat. The paper was shared 2 times in social media. The investigators investigate previously unseen data from Clio Online, the largest provider of digital learning content for primary schools in Denmark.
The paper shared the most on social media this week is by a team at Universidade Federal de Minas Gerais: "Auditing Radicalization Pathways on YouTube" by Manoel Horta Ribeiro et al (Aug 2019) with 3640 shares. The investigators conduct a large scale audit of user radicalization on YouTube. @DrengrV (Drengr) tweeted "Megan, imagine for a moment how scary it will be once the platforms that we use are created/controlled/hosted/owned by Us. What are you going to do then? You'll have no sway. No control over the pipeline. No data on the radicalization. It'll be completely out of your control.🐺❤".
This week was active for "Computer Science - Human-Computer Interaction", with 35 new papers.
This week was very active for "Computer Science - Learning", with 344 new papers.
The paper discussed most in the news over the past week was by a team at The Open University of Israel: "FSGAN: Subject Agnostic Face Swapping and Reenactment" by Yuval Nirkin et al (Aug 2019), which was referenced 5 times, including in the article New System Makes It Troublingly Easy to Create Deepfakes in Futurism. The paper got social media traction with 49 shares. On Twitter, @meverteam observed "Impressive advances in face swapping and reenactment by FSGAN", while @parand said "Deep fakes are going to get even more realistic".
Leading researcher Kyunghyun Cho (New York University) published "Latent-Variable Non-Autoregressive Neural Machine Translation with Deterministic Inference using a Delta Posterior".
The paper shared the most on social media this week is "Generating High-Resolution Fashion Model Images Wearing Custom Outfits" by Gökhan Yildirim et al (Aug 2019)
Over the past week, ten new papers were published in "Computer Science - Multiagent Systems".
Over the past week, 16 new papers were published in "Computer Science - Neural and Evolutionary Computing".
Over the past week, 35 new papers were published in "Computer Science - Robotics".
The paper discussed most in the news over the past week was "DoorGym: A Scalable Door Opening Environment And Baseline Agent" by Yusuke Urakami et al (Aug 2019), which was referenced 1 time, including in the article AI Stats News: 39% Of Business Executives Predict China Will Overtake US As The Global AI Leader in Forbes.com. The paper got social media traction with 43 shares. A user, @UrakamiYusuke, tweeted ""Domain randomization" + "realistic rendering" is one of the key to transfer the agent to the real world as OpenAI showed. But it wasn't easy to do it on your own. So we made the kit to do it!".