Deep neural networks have had an enormous impact on image analysis. State-of-the-art training methods, based on weight decay and DropOut, result in impressive performance when a very large training set is available. However, they tend to have large problems overfitting to small data sets. Indeed, the available regularization methods deal with the complexity of the network function only indirectly. In this paper, we study the feasibility of directly using the $L_2$ function norm for regularization. Two methods to integrate this new regularization in the stochastic backpropagation are proposed. Moreover, the convergence of these new algorithms is studied. We finally show that they outperform the state-of-the-art methods in the low sample regime on benchmark datasets (MNIST and CIFAR10). The obtained results demonstrate very clear improvement, especially in the context of small sample regimes with data laying in a low dimensional manifold. Source code of the method can be found at \url{https://github.com/AmalRT/DNN_Reg}.

Automatic image annotation has been an important research topic in facilitating large scale image management and retrieval. Existing methods focus on learning image-tag correlation or correlation between tags to improve annotation accuracy. However, most of these methods evaluate their performance using top-k retrieval performance, where k is fixed. Although such setting gives convenience for comparing different methods, it is not the natural way that humans annotate images. The number of annotated tags should depend on image contents. Inspired by the recent progress in machine translation and image captioning, we propose a novel Recurrent Image Annotator (RIA) model that forms image annotation task as a sequence generation problem so that RIA can natively predict the proper length of tags according to image contents. We evaluate the proposed model on various image annotation datasets. In addition to comparing our model with existing methods using the conventional top-k evaluation measures, we also provide our model as a high quality baseline for the arbitrary length image tagging task. Moreover, the results of our experiments show that the order of tags in training phase has a great impact on the final annotation performance.

Microsoft to turn millions of PCs and tablets into smart'home hubs' to take on Amazon's Echo and Google's Home Google's $130 Home speaker went on sale earlier this month Amazon's Alexa has been a huge hit with 5.1m sold Both can do everything from control lights to answer questions Microsoft plans to take them on with a new Home Hub app for Windows PCs Google's $130 Home speaker went on sale earlier this month Amazon's Alexa has been a huge hit with 5.1m sold The app will turn the millions of Windows PCs, laptops and tablets already in homes into a'home hub' that can do everything from control lights to answer questions. Has YOUR Google account been hacked? Researchers say... Apple goes Red for World AIDS day as firm is revealed to... Britain traded with the Middle East 1,300 years ago: Bitumen... The original human ancestor'Lucy' was a tree climbing... Has YOUR Google account been hacked? Researchers say... Apple goes Red for World AIDS day as firm is revealed to... Britain traded with the Middle East 1,300 years ago: Bitumen...

It's the home to North Korea's space program, and a cover for testing the technology that could one day send a nuclear-warhead-bearing missile hurtling towards the US mainland. Now thanks to some publicity photographs, modeling software, and the efforts of an enterprising video game developer, you can tour its command center without fear of arrest. The North Korea specialist website 38 North has released a 3-D model of the control room at the Sohae Satellite Launching Station, the building from which the North launches its satellites and test rocket engines. The model makes for a nifty virtual experience, showing viewers the heart of the facility in rich enough detail to spot the curvature of the rubber seals on the windows. You can see several of the renders in the gallery above.

Intelligence, defined as the ability to acquire knowledge and skills. Intelligence for the longest time possible is associated with the human brain. Artificial intelligence is basically defined as intelligence that is originating from machines. Most computer applications only make existing processes and functions faster and maybe more efficiently but cannot create new duties altogether. However, artificial intelligence has already challenged this notion.

Rob May is the CEO and Co-Founder of Talla, a platform for intelligent information delivery in Slack and Hipchat. Previously, Rob was the CEO and Co-Founder of Backupify, (acquired by Datto in 2014). Before that, he held engineering, business development, and management positions at various startups. Rob has a B.S. in Electrical Engineering and a MBA from the University of Kentucky. He is also a well-known angel investor in the AI space and is the creator and writer of the widely-read and highly-regarded AI newsletter, Technically Sentient. Rob May will be speaking at the Gigaom AI Now in San Francisco, February 15-16th. In anticipation of that, I caught up with him to ask a few questions.

Honda teases self-driving car concept with AI that can FEEL... Are iPhones now exploding? Apple users in China complain... The'Shazam for faces': World's first facial recognition for... Apple buys artificial intelligence firm that can read... Honda teases self-driving car concept with AI that can FEEL... Are iPhones now exploding? Apple users in China complain... The'Shazam for faces': World's first facial recognition for... Apple buys artificial intelligence firm that can read...

Instagram is making huge changes to comments, intended on making it a "welcoming place". The changes include the ability to stop comments from showing at all. It also lets people put little hearts next to any specific comment, the equivalent of liking them on Facebook. Previously, the ability to stop comments was a limited to a small number of high-profile accounts. But it will be made available to everyone in a few weeks, Instagram said.

Uber is now able to see where people are all the time – even if they've just finished their journey or aren't using the app at all. The app now asks for permission to see people's location always. That means that as long as the app has been opened and is on a phone, the company can see where its customers have been. Uber says that the feature has been added so that the app can see where people go just before and after they are being picked up. But it has also been described as "scary" by people who say that they don't want to use the app if it involves having their location known. In its facilities, JAXA develop satellites and analyse their observation data, train astronauts for utilization in the Japanese Experiment Module'Kibo' of the International Space Station (ISS) and develop launch vehicles 32/39 The robot developed by Seed Solutions sings and dances to the music during the Japan Robot Week 2016 at Tokyo Big Sight.

The US Supreme Court told a lower court to take another look at a $399 million award won by Apple from rival Samsung for copying the design of the iPhone. The unanimous decision extends a legal battle that dates back to 2011 and at one point spanned the globe and engulfed every major maker of smartphones. Writing for the court, Justice Sonia Sotomayor said Apple might not be entitled to Samsung's entire profit on 11 infringing smartphones. She told a federal appeals court to consider whether Apple should be able to recoup profits attributable only to particular components. The high court stopped short of deciding that question itself.