Artificial intelligence could scale new heights on its growing application in Southeast Asia, thanks to the region's relative openness to data collection. Bumrungrad International Hospital in Bangkok began using IBM's Watson AI platform to assist in cancer treatment late last year. The system automatically collects and learns from clinical studies and new papers, suggesting treatment options and giving expected success rates for each patient. Watson lets doctors make unbiased decisions on the best options, according to James Miser, Bumrungrad's chief medical information officer. The Thai hospital has consulted with Watson in treating breast, colon, lung and prostate cancer in about 50 patients so far and intends to expand its use.

Google's DeepMind division has announced a partnership with the NHS's Moorfields Eye Hospital to apply machine learning to spot common eye diseases earlier. The five-year research project will draw on one million anonymous eye scans which are held on Moorfields' patient database, with the aim to speed up the complex and time-consuming process of analysing eye scans. The hope is that this will allow diagnoses of common causes of sight loss, like diabetic retinopathy and age-related macular degeneration, to be spotted more rapidly and hence be treated more effectively. For example, Google says that up to 98 percent of sight loss resulting from diabetes can be prevented by early detection and treatment. Two million people are already living with sight loss in the UK, of whom around 360,000 are registered as blind or partially-sighted.

Toyota plans to have its self-driving car out by 2020. It's been testing a modified Lexus GS as well as their own hybrid self-driving vehicles on the road. In January 2016, Toyota announced the creation of the Toyota Research Institute (TRI), a 1 billion investment in AI to develop autonomous driving capabilities as well as home-care robots. Jim Adler, the first head of data at TRI, has been on the job for just two months. Before that, he was an executive at Metanautix, a data analytics platform that sold to Microsoft last year. Adler talked to TechRepublic about how Toyota is using data and simulation to teach cars to drive themselves. It sounded like so much fun and interesting, and leveraged quite a bit of my experience. How do you say "no" to working on a self-driving car and a robot?

Consider the following expenses claims: registration fees for a cancelled seminar, two separate claims for mileage when the employees travelled together, and a sandwich-and-coffee dinner claimed as the full per diem. While it's easy to believe that a few dishonest claims won't hurt, for individual victims, expenses fraud can be costly. Research conducted by the National Fraud Authority suggests that exaggerated expenses claims cost the British economy around 100 million annually; the private sector alone lost 80 million in 2013. Imagine if 20 per cent of your staff added 10 per cent to each mileage claim; the cumulative loss for the company would quickly become significant. Existing fraud detection systems flag dubious-looking expenses according to a set of rules, such as challenging claims in excess of a fixed cash amount or those that are 5 percent higher than claims submitted by peers in similar positions.

Machine learning techniques are often used in computer vision due to their ability to leverage large amounts of training data to improve performance. Unfortunately, most generic object trackers are still trained from scratch online and do not benefit from the large number of videos that are readily available for offline training. We propose a method for offline training of neural networks that can track novel objects at test-time at 100 fps. Our tracker is significantly faster than previous methods that use neural networks for tracking, which are typically very slow to run and not practical for real-time applications. Our tracker uses a simple feed-forward network with no online training required. The tracker learns a generic relationship between object motion and appearance and can be used to track novel objects that do not appear in the training set. We test our network on a standard tracking benchmark to demonstrate our tracker's state-of-the-art performance. Further, our performance improves as we add more videos to our offline training set. To the best of our knowledge, our tracker is the first neural-network tracker that learns to track generic objects at 100 fps.

In this paper, the online variants of the classical Frank-Wolfe algorithm are considered. We consider minimizing the regret with a stochastic cost. The online algorithms only require simple iterative updates and a non-adaptive step size rule, in contrast to the hybrid schemes commonly considered in the literature. Several new results are derived for convex and non-convex losses. With a strongly convex stochastic cost and when the optimal solution lies in the interior of the constraint set or the constraint set is a polytope, the regret bound and anytime optimality are shown to be ${\cal O}( \log^3 T / T )$ and ${\cal O}( \log^2 T / T)$, respectively, where $T$ is the number of rounds played. These results are based on an improved analysis on the stochastic Frank-Wolfe algorithms. Moreover, the online algorithms are shown to converge even when the loss is non-convex, i.e., the algorithms find a stationary point to the time-varying/stochastic loss at a rate of ${\cal O}(\sqrt{1/T})$. Numerical experiments on realistic data sets are presented to support our theoretical claims.

Heboi is a Japanese word that loosely means something is technically poor, or crappy. Thus: Hebocon, which, according to the founders, extends to both the robots and the people that make them -- but in an affectionate, pat-on-the-back kind of way. The competition involves several sumo-style matches in which the robots try to push their opponents out of the arena. It's no DARPA challenge, but when you see the robots in action you begin to realize it's almost as difficult a struggle. Most of the robots on show at Hebocon can't be controlled very well.

Who here remembers taking computer programming in school? Whether you learned programming by punching holes in a never ending series of cards, or by writing simple DOS or other computer language commands, the fact remained that computers needed an incredibly precise set of instructions to accomplish a task. The more complicated the task, the more complicated your instructions had to be. Machine learning is inherently different. Rather than telling a computer exactly how to solve a problem, the programmer instead tells it how to go about learning to solve the problem for itself.

Any aspiring science fiction writer looking for a good protagonist could do worse than ripping off the Wikipedia page for Demis Hassabis: He grew up in England as a chess prodigy and built absurdly sophisticated video games before getting a degree in computer science from Cambridge, started studying neuroscience and publishing respected papers on amnesia and other topics, and then proceeded to co-found one of the hottest artificial-intelligence startups. Now that his company, DeepMind, has been snapped up by Google for a reported 400 million to 500 million (depending on your tech blog of choice), exactly how this latest twist will change his story remains to be seen--but there's a decent chance Hassabis will ultimately become commander of an army of humanoid Googlebots. Google's acquisition of Hassabis and the rest of the DeepMind team points to the surging interest in the field of deep learning, a funky part of computer science seen as key to building truly intelligent machines. It centers on having computers learn to do tasks and find patterns on their own. Google, for example, received attention a couple of years ago, when its network of self-learning computers were able to understand the concept of a cat and find cats in YouTube videos.

Seymour Papert with LEGO Mindstorms robotics kits, which were named in recognition of Papert's seminal book, Mindstorms: Children, Computers, and Powerful Ideas. Seymour Papert with LEGO Mindstorms robotics kits, which were named in recognition of Papert's seminal book, Mindstorms: Children, Computers, and Powerful Ideas. The field of educational technology is mourning a visionary whose work was considered 50 years ahead of its time. Seymour Papert, who died July 31 at age 88, was a mathematician and computer scientist who spent decades at MIT. "Seymour was one of the very first people to recognize that new computer technologies could be used by kids to create things in new ways and express themselves," Mitchel Resnick, a professor of learning research at MIT and a longtime colleague and friend, told NPR Ed. "It's amazing that Seymour was thinking these ideas in the 1960s," Resnick adds, "when computers cost hundreds of thousands of dollars, but he foresaw the day that every child would have access to a computer." The great theme of Papert's work and life was the nature of intelligence, or what he called thinking about thinking.