News

Einstein’s ‘spooky action’ has just been demonstrated on a massive scale for the first time

Learning to Speak Shrub

A few months later, Karban, a professor at the University of California, Davis who studies plant defense communication, returns to the sagebrush and examines its leaves, many of which now have damage from real grasshoppers or beetles. However, within about two feet of the branches he clipped, leaves have been spared the worst ravages of the hungry insects. That’s because Karban’s cuttings convinced those damaged leaves they were under insect attack, so they sent chemical alarms into the air.

Measuring the Intrinsic Dimension of Objective Landscapes

In our paper, Measuring the Intrinsic Dimension of Objective Landscapes, to be presented at ICLR 2018, we contribute to this ongoing effort by developing a simple way of measuring a fundamental network property known as intrinsic dimension. In the paper, we develop intrinsic dimension as a quantification of the complexity of a model in a manner decoupled from its raw parameter count, and we provide a simple way of measuring this dimension using random projections. We find that many problems have smaller intrinsic dimension than one might suspect.

DeepMind papers at ICLR 2018

Between 30 April and 03 May, hundreds of researchers will gather in Vancouver, Canada, for the Sixth International Conference on Learning Representations. Here you will find details of all DeepMind’s accepted papers.

Forging Voices and Faces: The Dangers of Audio and Video Fabrication

NASA Mars Mission Faces Setback After Heat Shield Cracks Under Pressure

Algorithms tutorial for beginners

As a developer, you have the power to change the world! You can write programs that enable new technologies. You might work in software to find an earlier diagnosis of diseases.

Pentagon-funded research aims to predict when crimes are gang-related

The paper attempts to predict whether crimes are gang-related using a neural network, a complex computational system modeled after a human brain that “learns” to classify or identify items based on ingesting a training dataset. The authors selected what they determined to be the four most important features (number of suspects, primary weapon used, the type of premises where the crime took place, and the narrative description of the crime) for identifying a gang-related crime from 2014–16 LAPD data and cross-referenced the crime incidents with a 2009 LAPD map of gang territory to create a training dataset for their neural network.

Another Side of Feynman

Yesterday I went for a long walk in the spring sunshine with Trudy Eyges and Richard Feynman. Feynman is the young American professor, half genius and half buffoon, who keeps all physicists and their children amused with his effervescent vitality. He has, however, as I have recently learned, a great deal more to him than that, and you may be interested in his story.

TDM: From Model-Free to Model-Based Deep Reinforcement Learning

While simple, this thought experiment highlights some important aspects of human intelligence. For some tasks, we use a trial-and-error approach, and for others we use a planning approach. A similar phenomena seems to have emerged in reinforcement learning (RL).