Second-order optimization methods such as natural gradient descent have the potential to speed up training of neural networks by correcting for the curvature of the loss function. Unfortunately, the exact natural gradient is impractical to compute for large models, and most approximations either require an expensive iterative procedure or make crude approximations to the curvature. We present Kronecker Factors for Convolution (KFC), a tractable approximation to the Fisher matrix for convolutional networks based on a structured probabilistic model for the distribution over backpropagated derivatives. Similarly to the recently proposed Kronecker-Factored Approximate Curvature (K-FAC), each block of the approximate Fisher matrix decomposes as the Kronecker product of small matrices, allowing for efficient inversion. KFC captures important curvature information while still yielding comparably efficient updates to stochastic gradient descent (SGD). We show that the updates are invariant to commonly used reparameterizations, such as centering of the activations. In our experiments, approximate natural gradient descent with KFC was able to train convolutional networks several times faster than carefully tuned SGD. Furthermore, it was able to train the networks in 10-20 times fewer iterations than SGD, suggesting its potential applicability in a distributed setting.

I love the feeling of having a new way to think about the world. I especially love when there's some vague idea that gets formalized into a concrete concept. Information theory is a prime example of this. Information theory gives us precise language for describing a lot of things. How uncertain am I? How much does knowing the answer to question A tell me about the answer to question B? How similar is one set of beliefs to another? I've had informal versions of these ideas since I was a young child, but information theory crystallizes them into precise, powerful ideas. These ideas have an enormous variety of applications, from the compression of data, to quantum physics, to machine learning, and vast fields in between. Unfortunately, information theory can seem kind of intimidating. I don't think there's any reason it should be. In fact, many core ideas can be explained completely visually! Before we dive into information theory, let's think about how we can visualize simple probability distributions. We'll need this later on, and it's convenient to address now. As a bonus, these tricks for visualizing probability are pretty useful in and of themselves! Sometimes it rains, but mostly there's sun! Let's say it's sunny 75% of the time. It's easy to make a picture of that: Most days, I wear a t-shirt, but some days I wear a coat. Let's say I wear a coat 38% of the time. It's also easy to make a picture for that! What if I want to visualize both at the same time?

I can't remember the last time I called someone on the phone to place an order for pizza. Papa John's told investors during its first-quarter earnings call that 55% of its total sales now come through digital. Sixty percent of those digital transactions came from mobile devices. Competing pizza chains Domino's and Yum! Brands' Pizza Hut are also investing heavily in e-commerce. Both have invested in ordering methods for smartwatches, connected cars, and video game systems.

LAGOS--Nigeria is beating the West at its own word game, using a strategy that sounds like Scrabble sacrilege. By relentlessly studying short words, this country of 500 languages has risen to dominate English's top lexical contest. Last November, for the final of Scrabble's 32-round World Championship in Australia, Nigeria's winningest wordsmith, Wellington Jighere, defeated Britain's Lewis Mackay, in a victory that led morning news broadcasts in his homeland half a world away. It was the crowning achievement for a nation that boasts more top-200 Scrabble players than any other country, including the U.K., Nigeria's former colonizer and one of the board game's legacy powers. "In other countries they see it as a game," said Mr. Jighere, now a borderline celebrity and talent scout for one of the world's few government-backed national programs.

Brian Christian is the author of The Most Human Human, which was named a Wall Street Journal bestseller and a New Yorker favourite book of 2011, and has been translated into ten languages. He is the coauthor, with Tom Griffiths, of Algorithms to Live By, available Spring 2016. The Most Human Human is a provocative, exuberant, and profound exploration of the ways in which computers are reshaping our ideas of what it means to be human. He has just released his latest book, Algorithms to Live By. Co-written with cognitive scientist Tom Griffiths, the book offers a fascinating exploration of how computer algorithms can be applied to our everyday lives, helping to solve common decision-making problems and illuminate the workings of the human mind.

Everywhere you turn these days there are more and more automated processes appearing all the time. From automatic vacuum cleaners to self-order counters at restaurants, to cars that automatically park themselves, robots are all around us in one way or another and physics is no different. In using the latest artificial intelligence to do the same tasks as people, we are not only saving time and money but saving on resources too. A recent physics experiment developed by physicists from The Australian National University (ANU) and the University of New South Wales at the Australian Defence Force Academy (UNSW ADFA) was shown to be completed by artificial intelligence (AI) just as a human would. The test was to create a replica of "Laser Beam" experiment that won the 2001 Nobel Prize and produced an extremely cold gas trapped in a laser beam (known as Bose-Einstein condensate) and the incredible AI literally taught itself how to do the experiment, from start to finish, in under one hour!

Professor Damith Herath will speak about artificial intelligence and robotics at King O'Malley's Pub on May 25 as part of Canberra's first Pint of Science series. Canberra will raise a glass to a fresh take on politics in the pub this week as the Pint of Science festival debuts in the capital. The national series, funded by the Australian Department of Industry, Innovation and Science and the ACT government, brings some of the most brilliant scientists into city pubs to discuss their latest research and field questions about their findings. The action will begin on Monday at King O'Malley's in Civic. For just 5 punters can delve into how we learn and the power of the brain at Beautiful Mind from 7pm. At the same time on Tuesday evening ANU professor of physics Daniel Shaddock and ANU vice-chancellor Brian Schmidt will explain how we listen to and measure things in space at Ever Expanding Gravitational Pull.

Canberra will raise a glass to a fresh take on politics in the pub this week as the Pint of Science festival debuts in the capital. The national series, funded by the Australian Department of Industry, Innovation and Science and the ACT government, brings some of the most brilliant scientists into city pubs to discuss their latest research and field questions about their findings. The action will begin on Monday at King O'Malley's in Civic. For just 5 punters can delve into how we learn and the power of the brain at Beautiful Mind from 7pm. At the same time on Tuesday evening ANU professor of physics Daniel Shaddock and ANU vice-chancellor Brian Schmidt will explain how we listen to and measure things in space at Ever Expanding Gravitational Pull. And the finale event, Driven Into Crazy Stupid Love, on Wednesday from 7pm will cover self-driving cars and what the world will be like with increased human-robot interaction.

An early Google employee tests different apps in his recently remodeled San Francisco home. The home's automation system can be accessed by smart phone; the front door has an automated video doorbell. Facebook founder and chief Mark Zuckerberg created the world's largest media owner, with 1.6 billion users, but it does not produce content. DIGITAL technology has quickly knocked some of the world's biggest businesses off their perches and will soon make more radical changes to our everyday lives. Tech experts have been surprised by the speed that consumers have flocked to global online platforms such as Uber and Airbnb, and believe it's only the beginning -- with artificial intelligence and internet-connected homes among the next big growth areas.

Human communication goes beyond words. It is complex, rich in nuances and frequently includes non-verbal signs. Yet despite our technological limitations it is not impossible for some aspects of communication to be emulated by a machine with surprising effect. This has been part of the challenge in developing Harlie (Human and Robot Language Interaction Experiment), a smartphone chatbot app developed by researchers at the CSIRO and University of Queensland. It's primarily aimed at people who may have trouble conversing including those with neurological conditions such as Parkinson's disease and dementia, or even autism.