In March, Microsoft revealed that it was using the open-world game Minecraft to train AI agents to learn how to do things like climbing a hill. The company also promised to make it available to the public so they could work on their own artificial intelligence projects and research, and it's finally available today. Project Malmo (formerly known as Project AIX) is a Minecraft mod that works on Windows, Mac and Linux, and supports just about any programming language you might want to use. So yes, that means you will need to know how to code – but Microsoft says that even novice programmers can get in on the action. You can learn more about Project Malmo here and grab the mod from this GitHub repository to try it for yourself.

Red Dead Redemption arrives on Xbox One, ID@Xbox 100 million success stories, Battlefield 1's destruction and more! Every week, I summarize some of the biggest rumors and talking points in one handy post every weekend. You can also rely on this weekly column to catch up with all the latest Xbox One game releases. Welcome to the latest edition of This Week in Xbox One News. Red Dead Redemption hit backwards compatibility this week, causing a huge sales spike for the game on Amazon (via ICXM).

New Delhi: Aiming to attract more fliers and achieve higher seat occupancy in its flights on trunk routes, Air India has decided to lower its last minute ticket fares to the level of 2-tier AC of Rajdhani trains. As part of the pricing strategy, Air India will drop fares on four key routes--Delhi-Mumbai, Delhi-Chennai, Delhi-Kolkata and Delhi-Bengaluru--four hours before the departure of flights to these destinations. Air India Chairman and Managing Director Ashwani Lohani said the move is aimed at providing relief to the passengers from last minute sky-rocketing fares and also to fill the vacant seats. The fare of second AC Rajdhani ticket for Delhi-Kolkata is Rs 2,890 and Rs 4,095 for Delhi-Bengaluru Rajdhani. Air India has an average load factor of 74 per cent across its domestic network while the seat occupancy on these trunk routes stands at around 80 per cent, he said.

We consider the problem of approximating partition functions for Ising models. We make use of recent tools in combinatorial optimization: the Sherali-Adams and Lasserre convex programming hierarchies, in combination with variational methods to get algorithms for calculating partition functions in these families. These techniques give new, non-trivial approximation guarantees for the partition function beyond the regime of correlation decay. They also generalize some classical results from statistical physics about the Curie-Weiss ferromagnetic Ising model, as well as provide a partition function counterpart of classical results about max-cut on dense graphs \cite{arora1995polynomial}. With this, we connect techniques from two apparently disparate research areas -- optimization and counting/partition function approximations. (i.e. \#-P type of problems). Furthermore, we design to the best of our knowledge the first provable, convex variational methods. Though in the literature there are a host of convex versions of variational methods \cite{wainwright2003tree, wainwright2005new, heskes2006convexity, meshi2009convexifying}, they come with no guarantees (apart from some extremely special cases, like e.g. the graph has a single cycle \cite{weiss2000correctness}). We consider dense and low threshold rank graphs, and interestingly, the reason our approach works on these types of graphs is because local correlations propagate to global correlations -- completely the opposite of algorithms based on correlation decay. In the process we design novel entropy approximations based on the low-order moments of a distribution. Our proof techniques are very simple and generic, and likely to be applicable to many other settings other than Ising models.

The minimum description length (MDL) principle in supervised learning is studied. One of the most important theories for the MDL principle is Barron and Cover's theory (BC theory), which gives a mathematical justification of the MDL principle. The original BC theory, however, can be applied to supervised learning only approximately and limitedly. Though Barron et al. recently succeeded in removing a similar approximation in case of unsupervised learning, their idea cannot be essentially applied to supervised learning in general. To overcome this issue, an extension of BC theory to supervised learning is proposed. The derived risk bound has several advantages inherited from the original BC theory. First, the risk bound holds for finite sample size. Second, it requires remarkably few assumptions. Third, the risk bound has a form of redundancy of the two-stage code for the MDL procedure. Hence, the proposed extension gives a mathematical justification of the MDL principle to supervised learning like the original BC theory. As an important example of application, new risk and (probabilistic) regret bounds of lasso with random design are derived. The derived risk bound holds for any finite sample size $n$ and feature number $p$ even if $n\ll p$ without boundedness of features in contrast to the past work. Behavior of the regret bound is investigated by numerical simulations. We believe that this is the first extension of BC theory to general supervised learning with random design without approximation.

We consider a collaborative online learning paradigm, wherein a group of agents connected through a social network are engaged in playing a stochastic multi-armed bandit game. Each time an agent takes an action, the corresponding reward is instantaneously observed by the agent, as well as its neighbours in the social network. We perform a regret analysis of various policies in this collaborative learning setting. A key finding of this paper is that natural extensions of widely-studied single agent learning policies to the network setting need not perform well in terms of regret. In particular, we identify a class of non-altruistic and individually consistent policies, and argue by deriving regret lower bounds that they are liable to suffer a large regret in the networked setting. We also show that the learning performance can be substantially improved if the agents exploit the structure of the network, and develop a simple learning algorithm based on dominating sets of the network. Specifically, we first consider a star network, which is a common motif in hierarchical social networks, and show analytically that the hub agent can be used as an information sink to expedite learning and improve the overall regret. We also derive networkwide regret bounds for the algorithm applied to general networks. We conduct numerical experiments on a variety of networks to corroborate our analytical results.

Machine learning artificial intelligence has unlocked big data as a source of military, weather and business intelligence that has opened up multiple options. Social Media giants Twitter and Facebook have been spending millions trying to keep their companies ahead of the flock, highlighted by Twitter Buys Machine Learning Artificial Intelligence Star Magic Pony Technology Pavel Machalek co-founder of Silicon Valley data analytics firm Spaceknow working with commercial satellite data says the convergence of computing power, machine learning and satellite imagery is a perfect storm that s just beginning to peak, ... We could not have done this five years ago. Chinese government economic reports are notoriously inaccurate. Spaceknow's China Satellite Manufacturing Index uses satellite imagery to monitor changes at 6,000 industrial facilities in China as an alternative. The above image courtesy of DigitalGlobe shows how geospatial data companies can track activity by identifying surface material as seen here with individual trees in a forest (above) and aircrafts on the tarmac (below).

Artificial intelligence (AI) is intelligence created by machines or software. The artificial intelligence and robotics market was worth US 10.7 billion in 2014 and is expected to be worth US 153 billion by 2020, and to have a disruptive impact of between US 14 to US 33 trillion. The component for artificial intelligence alone is worth US 70 billion. There are perils with artificial intelligence. Don't let anyone tell you there aren't.

Top Machine Learning & Data Mining Books - in this post, we have scraped various signals (e.g. We have combined all signals to compute the Quality Score for each book and publish the list of top Machine Learning and Data Mining books. The readers will love the list because it is data-driven & objective. This book is very well rated on Amazon website and is written by three professors from USC, Stanford and University of Washington. The book's authors: Gareth James, Daniela Witten, Trevor Hastie, & Rob Tibshirani all have backgrounds in statistics.

By Riki Ozawa / Yomiuri Shimbun Staff WriterAutomated cars are classified into four levels based on how much the driver is involved in the operation of the accelerator, brakes and steering. The ranks range from Level 1 -- in which one of the operations is automated -- to Level 4 -- in which a car is fully autonomous and can operate without a human driver. Most domestic and overseas car manufacturers are at Level 2, with Tesla Motors Inc.'s Model S also in this category. Toyota Motor Corp., Nissan Motor Co., Honda Motor Co. and other major domestic carmakers have been putting work into the research and development process, aiming to market automated cars that can run on expressways by around 2020. The manufacturers are aiming to reach Level 3, in which a self-driving system mainly operates the car and the driver takes the necessary actions only in case of emergency, such as when the system malfunctions.