When using reinforcement learning (RL) algorithms to evaluate a policy it is common, given a large state space, to introduce some form of approximation architecture for the value function (VF). The exact form of this architecture can have a significant effect on the accuracy of the VF estimate, however, and determining a suitable approximation architecture can often be a highly complex task. Consequently there is a large amount of interest in the potential for allowing RL algorithms to adaptively generate approximation architectures. We investigate a method of adapting approximation architectures which uses feedback regarding the frequency with which an agent has visited certain states to guide which areas of the state space to approximate with greater detail. This method is "unsupervised" in the sense that it makes no direct reference to reward or the VF estimate. We introduce an algorithm based upon this idea which adapts a state aggregation approximation architecture on-line. A common method of scoring a VF estimate is to weight the squared Bellman error of each state-action by the probability of that state-action occurring. Adopting this scoring method, and assuming $S$ states, we demonstrate theoretically that - provided (1) the number of cells $X$ in the state aggregation architecture is of order $\sqrt{S}\log_2{S}\ln{S}$ or greater, (2) the policy and transition function are close to deterministic, and (3) the prior for the transition function is uniformly distributed - our algorithm, used in conjunction with a suitable RL algorithm, can guarantee a score which is arbitrarily close to zero as $S$ becomes large. It is able to do this despite having only $O(X \log_2S)$ space complexity and negligible time complexity. The results take advantage of certain properties of the stationary distributions of Markov chains.

While many firms are looking to the moon for mining opportunities, one Australian firm believes there could be precious metals a lot nearer to home. Deep-sea robots will be sent to mine mineral deposits in the deep ocean in 2019 in a test for a controversial new scheme. As land-based mineral stores are becoming depleted, the ocean floor is becoming a more attractive mining prospect, containing gold, copper and other precious metal deposits used to make electronics, renewable energy tools and even medical imaging machines. But deep-sea excavation may have a negative impact on deep ocean marine life, as robot mining may destroy their homes and disturb these sensitive species. The Canadian mining company Nautilus Minerals plans to send robots to mine deposits rich in copper and gold in the waters of Papua New Guinea.

Domino's latest technical achievement may well provoke a sense of déjà vu in many customers who used to order pizza by phone. The next time you place an order on the Domino's app or website, you'll be able to talk to a robot that will take the order. DRU Assist is a virtual assistant developed by Domino's in conjunction with voice technology provider Nuance Communications. "DRU Assist is not just a toy. We genuinely see this as a real platform change," Domino's CEO Don Meij said at the company's regular technology showcase in Sydney today.

Domino's has delivered new additions to its Domino's Robotics Unit (DRU) family, announcing the DRU Platform, an artificial intelligence (AI)-based technology that will allow customers to order a pizza using their voice. As of Monday, customers will be able to use DRU Assist, an in-app AI virtual assistant that was built in partnership with natural language company Nuance. DRU Assist takes the pizza giant somewhat back to the start of its digital journey, with customers speaking to order a pizza via their phone. Powered by Nuance's Nina, Domino's DRU Assist engages with customers in human-like conversation via text or speech recognition. "DRU Assist is not just a toy, this is real platform change," Domino's Group CEO and Managing Director Don Meij said.

Microsoft Monday is a weekly column that focuses on updates in regards to the Redmond giant. This week, Microsoft Monday includes details about Windows 10 upgrade revealed, EU still not satisfied with data collection transparency, HoloLens used for designing operating rooms, the launch of Skype Lite, an investment in AirMap, Halo split-screen is returning, new features for the Mail and Calendar apps for Windows 10 and much more! Microsoft has recently announced that a fourth major Windows 10 upgrade is being planned. Microsoft made the announcement at Ignite Australia roughly about two weeks ago. The first three updates are known as November Update, Anniversary Update and Creators Update.

In my last blog, I discussed how digital complacency could put your business at risk and how technological change is fast outpacing businesses ability to adapt. This trend continues its relentless pursuit to consume those whom choose to answer today's questions with yesterday's thinking. Furthermore, as we move into 2017 yet another transformational technology is redefining the industry and it has the potential to change everything. Machine Learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. Machine learning focuses on the development of computer programs that can teach themselves to grow and change when exposed to new data.

A cosmic jellyfish floats through the sea in a new video from an underwater robotic craft tasked with exploring the deep ocean off the coast of American Samoa and Samoa. The Okeanos Explorer ship -- operated by the National Oceanic and Atmospheric Administration (NOAA) -- caught sight of the jellyfish on Feb. 21 during a dive in the Utu seamount, a previously unexplored area, according to the agency. The researchers who caught sight of the jellyfish noted that it has two sets of tentacles and the yellow glowing structures dotting the outside of the jelly appear to be reproductive organs, with its red digestive system in the middle. "It's a really beautiful jellyfish," one of the scientists said on a NOAA website. The jellyfish wasn't the only creature seen during the dive deep into Utu.

More artificial intelligence, cheaper sensors and longer flying drones are only some of the technological advances that Kiwi farmers can look forward to on "data-driven" farms over the next 10 years. Microsoft Research principal researcher Ranveer Chandra has been in New Zealand for a week offering insights into precision agriculture and advances the United States technology company is working on to improve farming and food production. He said Kiwi farmers and their innovations could help lead world farming, but they would see more advances themselves over the next decade. Ranveer Chandra is a principal researcher for Microsoft. Farmers faced doubling food production to feed a growing population by 2050 and this would require more technological advances world-wide, he said..

Current research projects show that artificial intelligence (AI) can help us solving global problems for the greater good of mankind. Despite the headlines presenting crashes involving self-driving cars, the applications of AI technology in this area could dramatically reduce injuries and deaths on the roads. Self-driving cars could also change our lifestyles. Thanks to this emerging technology, we will have more time to entertain ourselves of work during commutes. The decreased cognitive load and the increased comfort provided by shared transportation and self-driving may also affect where people choose to live.

You no longer need a Pixel to use Google Assistant on your smartphone. Google is finally expanding its Assistant to phones beyond the Pixel line. The company is starting to roll out the intelligent assistant feature to nearly all phones running Android Marshmallow and Nougat (some low-end devices excepted.) SEE ALSO: Hey, Siri: How'd you and every other digital assistant get its name? The update, which will be part of a change to the Google app, will be available in the U.S. beginning this week and will then launch in the UK, Australia, Canada and Germany.