tag > ML
-
A Probabilistic Formulation of Unsupervised Text Style Transfer - ICLR2020 paper by Junxian He, Xinyi Wang, Graham Neubig, Taylor Berg-Kirkpatrick
A deep generative model for unsupervised text style transfer, by hypothesizing a parallel latent (unobserved) sequence to generate each observed sequence, good results on benchmarks.
-
ARIES - Artificial Intelligence for Ecosystem Services
ARIES is a networked software technology that redefines ecosystem service assessment and valuation for decision-making. The ARIES approach to mapping natural capital, natural processes, human beneficiaries, and service flows to society is a powerful new way to visualize, value, and manage the ecosystems on which the human economy and well-being depend.
-
Growing Neural Cellular Automata - Differentiable Model of Morphogenesis - by Alexander Mordvintsev, Ettore Randazzo, Eyvind Niklasson, Michael Levin
-
John Horton Conway: "On His LOVE/HATE Relationship with LIFE"
-
AI contributes to coronavirus diagnosis (China Daily)
The Mudanyuan station on the Beijing subway's Line 10 uses an artificial intelligence based fever screening system. Developed by Beijing-based Megvii and put into use on Tuesday, it supports non-contact remote temperature screening from a distance of 3 meters. The system's margin of error is within 0.3 degree Celsius and it can send fever alerts for up to 15 people per second.
-
Will Quantum Supremacy Start the Modern-Day Industrial Revolution? - talk by Hartmut Neven (Engineering Director at Google) - at Google Zeitgeist 2019
-
Modern manufacturing’s triple play: Digital twins, analytics & the internet of things (Economist)
Imagine if doctors could create a digital twin version of you that updates in real time using data from sensors in your home, in your car and on your person, such as wearable devices like smart watches. This virtual twin could signal or even predict impending illnesses such as cancer, thus enabling early diagnosis when treatment might be most effective.
-
Interview with Dr. Saleema Amershi on Responsible AI
-
Climate Models Are Running Red Hot, and Scientists Don’t Know Why (Bloomberg) - "The simulators used to forecast warming have suddenly started giving us less time."
-
Unsupervised Doodling and Painting with Improved SPIRAL (paper)
We investigate using reinforcement learning agents as generative models of images. A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent’s samples, either unconditional or reconstructions.
-
Differentiable Digital Signal Processing (DDSP) (code)
Fusing classic interpretable DSP with neural networks.
-
Google says its new chatbot Meena is the best in the world (MIT Tech Review)
Google.AI: In “Towards a Human-like Open-Domain Chatbot”, we present Meena, a 2.6 billion parameter end-to-end trained neural conversational model that learns to respond sensibly to a given conversational context. We show that Meena can conduct conversations that are more sensible and specific than existing state-of-the-art chatbots.
-
Quantitative spatial upscaling of categorical information: The multi‐dimensional grid‐point scaling algorithm - by Daniel Gann (2019)
A new scaling algorithm for categorical data generates representative classification system and reduces information loss.
