The biggest night in Hollywood is around the corner, and with it comes the possibility for a fresh batch of records to be broken and history to be made. Below, a few of the biggest Oscar milestones that may come to fruition on Sunday night. Outside of the technical categories, sci-fi movies have always had a tough go at the Academy Awards. With rare exception, they're usually overlooked for Best Picture--but depending upon who you ask, Guillermo del Toro's Creature From the Black Lagoon–inspired love story between a woman and a fish-man is the current front-runner (or close to it). Though their respective films, 12 Years a Slave and Moonlight, were crowned Best Picture, Steve McQueen and Barry Jenkins didn't manage to take home the coveted directing award.

The very idea of the first camera from Google causes people to freak out. The tiny Clips camera, which arrived in stores Tuesday, decides on its own what to record. Whenever it's on, artificial intelligence inside is watching. "This doesn't even *seem* innocent," tweeted Elon Musk after Google announced Clips in October. When I first saw one, I couldn't shake the feeling Google had created the creepy surveillance contraption in Dave Eggers' Silicon Valley horror "The Circle."

Today's best Amazon deals are on TVs, robot vacuums, cookware, and more. If you make a purchase by clicking one of our links, we may earn a small share of the revenue. However, our picks and opinions are independent from USA Today's newsroom and any business incentives. Sometimes Amazon's best deals serve up opportunities to pick up things that have been on your wish list for ages, and sometimes you can find a bunch of low-cost items that are perfect to scratch the online shopping itch. Today's top offers fall into the former category.

Sky has revealed a whole range of new updates to its Q boxes. The range of new features and tweaks will be arriving on viewers' TVs over the coming months. They include features like Spotify on the TV box, a doubling of the amount of the 4K content that will be available, improved voice functionality, and greater personalisation. The company will also change the design of the menus that are used to get around, including a new widescreen interface and a special mode for kids. Spotify is perhaps the most significant update, marking a major upgrade in the way that people can listen to music on their Sky box.

All of a sudden, Sky has become the prize in a bidding war between US media giants Fox and Comcast, but for now, it's business as usual on the ground. The pay-TV provider has today announced a number of new features coming to its Sky Q service, primarily focused on making the user experience more personalised. Machine learning will begin to play a more important role throughout -- surfacing different TV shows depending on the time of day, for example. The sports tab is also becoming better tailored, assessing your viewing habits and liaising with the Sky Sports app so it knows your favourite events and teams. These updates will feed into a "brand new widescreen user interface" that blends personal recommendations with a selection of on-demand TV curated by Sky.

In the last years there has been quite some fuzz in the news about two seemingly unrelated topics: the decrease of bees and other pollinators and, simultaneously, the increase of artificial intelligence and data science. Reading up a bit on pollination and how research on pollination works I realized that although a lot is known, there is, as in all sciences, way more unknown. This triggered my curiosity, wouldn't it be possible to use AI for the study of pollinators? In this post I explain a simple way to do so. I wrote some code, which can be found om my github. But as a teaser: I wanted to quantify the number of visitations to the flowers in this movie.

Artificial intelligence is breaking out of science fiction and sprinting into reality. A robot can now identify a human from a photo of their face, trounce a Poker player and in theory, pilot a plane. Business applications for AI are growing, from Apple's Siri personal assistant to Amazon's delivery drones, and business schools want to ensure that their graduates have the skills to meet the future needs of industry. A wide range of master's degrees and electives within established courses that focus on AI are on offer. "Digital and AI are rapidly changing the way we live and work in significant ways," says Francisco Veloso, dean of London's Imperial College Business School. As businesses transform to keep up with the pace of technological change, he adds, "schools will need to do more to provide students with the tools they need to undertake careers or start businesses in areas such as blockchain, [and] fintech".

The first interactor--a muscular man in his fifties with a shaved head and a black V-neck sweater--walks into a conference room and sits in a low-slung blue armchair before a phalanx of video cameras and studio lights. The rest of the room is totally dark. He gazes at a black, hockey- puck-shaped object--an Amazon Echo--on a small table in front of him. "Alexa," he says, "let's chat." "Good morning, my friend," a female voice replies with synthetic agreeability, a purplish ring of light pulsing atop the Echo.

With the launch of NytDEMO, the Times looks to help ease one of marketers' top challenges in the digital era: targeting the right audience with the right message at the right time. Consumers increasingly crave personalization from brands, but marketers struggle with meeting these demands. A lack of personalization cost businesses as much $756 billion in 2016 and 41% of consumers reported switching companies due to lack of personalization, according to Accenture. An additional 31% of respondents said they find great value in services that automatically learn about their needs and tailor recommendations accordingly. AI-driven solutions provide more reliable insights that marketers can use to achieve higher levels of personalization, create higher-quality content and deploy more transparent campaigns.

I'm playing with CapsNet using Tensorflow to classify images taken from a traffic cam. I decided to start with just two classes (Fiat-Panda and Fiat-500) and I created a dataset with cropped images of car's rear. I reached a 0.56 accuracy on the testset which is pretty low for a 2-class classification problem. Here you can see the recap. The decoder on the final caps layer outputs mixed images of the two classes and the loss function on the test set doesn't converge.