Soren Schwertfeger finished his postdoctorate research on autonomous robots in Germany and seemed set to continue his work in Europe or the United States, where artificial intelligence was pioneered and established. Instead, he went to China. "You couldn't have started a lab like mine elsewhere," Schwertfeger said. The balance of power in technology is shifting. China, which for years watched enviously as the West invented the software and the chips powering today's digital age, has become a major player in artificial intelligence, what some think may be the most important technology of the future.

The incorporation of prior knowledge into learning is essential in achieving good performance based on small noisy samples. Such knowledge is often incorporated through the availability of related data arising from domains and tasks similar to the one of current interest. Ideally one would like to allow both the data for the current task and for previous related tasks to self-organize the learning system in such a way that commonalities and differences between the tasks are learned in a data-driven fashion. We develop a framework for learning multiple tasks simultaneously, based on sharing features that are common to all tasks, achieved through the use of a modular deep feedforward neural network consisting of shared branches, dealing with the common features of all tasks, and private branches, learning the specific unique aspects of each task. Once an appropriate weight sharing architecture has been established, learning takes place through standard algorithms for feedforward networks, e.g., stochastic gradient descent and its variations. The method deals with domain adaptation and multi-task learning in a unified fashion, and can easily deal with data arising from different types of sources. Numerical experiments demonstrate the effectiveness of learning in domain adaptation and transfer learning setups, and provide evidence for the flexible and task-oriented representations arising in the network.

In this paper, we extend the geometric descent method recently proposed by Bubeck, Lee and Singh to tackle nonsmooth and strongly convex composite problems. We prove that our proposed algorithm, dubbed geometric proximal gradient method (GeoPG), converges with a linear rate $(1-1/\sqrt{\kappa})$ and thus achieves the optimal rate among first-order methods, where $\kappa$ is the condition number of the problem. Numerical results on linear regression and logistic regression with elastic net regularization show that GeoPG compares favorably with Nesterov's accelerated proximal gradient method, especially when the problem is ill-conditioned.

Every six months, an enormous posse of top robotics researchers from around the world converge on some moderately exotic location to impress each other with their latest research. Right now, we're at the 2017 edition of the IEEE International Conference on Robotics and Automation (ICRA), which is taking place as you read this in Singapore. As always, we're going to do our best to read every single paper and attend every single technical session, even though there are 11 tracks all happening at the same time along with workshops, forums, and an expo. You can expect to see posts this week about what's most novel and interesting from the conference, but (again, as always) there's way too much cool stuff to cram into just one week: We'll continue to post ICRA-related content for the next few weeks, even as we get back to our regular robotics news. And if you're at ICRA, come say hi! I'll be the harassed-looking dude with a big camera frantically running between sessions.

Anticipation is growing for the July release of Tesla's'affordable' $35,000 Model 3. Now, the latest shots of Telsa testing the car may have revealed one of its final secrets - what the interior will look like. The shots, taken near Tesla's headquarters in Palo Alto, California, reveal the car has a single screen and no traditional instruments. Musk had previously addressed the issue of no traditional dashboard display, tell users who asked for one'You won't care' before confirming the car won't use a heads up display either. A new comparison of models on Tesla's own site confirms the spec, saying the Model 3 will only have a single 15inch display, while the more expensive Model S has a separate'driver display'. According to Teslerati, 'The latest photos gives us a clearer look at the landscape-mounted touchscreen, which resembles an off-the-shelf computer monitor that's been bolted onto the dashboard.'

From in-depth features and interviews to the daily torrent of trailers and news, we write a lot about video games here. But there's only so much one team can cover, and often some of our favorite games never grace the digital pages of Engadget. To remedy that shortcoming, we're introducing Gaming IRL, a monthly segment where several editors talk about what they've been playing in their downtime. Sometimes these'll be the latest AAA game, but you'll also see free-to-play mobile distraction and classics revisited (or criminally ignored until now). Gaming IRL is part of a broader series in which you'll find stories from all of the areas we cover: gadgets we use every day, the apps and services we adore, what we're watching and the music and podcasts we can't live without.

AlphaGo is going out on top. After beating Ke Jie, the world's best player of the ancient Chinese board game Go, for the third time today at the Future of Go Summit in Wuzhen, Google's DeepMind unit announced that it would be the last event match the AI plays. In a statement, DeepMind co-founder and co-CEO Demis Hassabis said the reason was that this week's summit represented "the highest possible pinnacle for AlphaGo as a competitive program." AlphaGo rose to prominence a little over a year ago when it unexpectedly defeated legendary player Lee Se-dol 4-1 in a match held in Seoul. Most computer scientists expected the feat of beating a top Go player with artificial intelligence to be decades away due to the game's complexity and nuance, but with this week's comprehensive defeat of Ke Jie the matter has been settled.

On the eve of Computex, Taiwan's big showpiece event where PC makers roll out the latest and best implementations of Intel CPUs, mobile rival ARM is announcing its own big news with the unveiling of a new generation of ARM CPUs and GPUs. Official today, the ARM Cortex-A75 is the new flagship-tier mobile processor design, with a claimed 22 percent improvement in performance over the incumbent A73. It's joined by the new Cortex-A55, which has the highest power efficiency of any mid-range CPU ARM's ever designed, and the Mali-G72 graphics processor, which also comes with a 25 percent improvement in efficiency relative to its predecessor G71. The efficiency improvements are evolutionary and predictable, but the revolutionary aspects of this new lineup relate to artificial intelligence: this is the first set of processing components designed specifically to tackle the challenges of onboard AI and machine learning. Plus, last year's updates to improve performance in the power-hugry tasks of augmented and virtual reality are being extended and elaborated.

Hundreds of AI researchers have taken a glimpse into their crystal balls to try to determine when machines will finally exceed human capabilities. A survey run by the Future of Humanity Institute, a research center that studies existential risks at the University of Oxford in the UK and Yale University in the US, asked 352 machine learning researchers to predict how AI will progress. "Researchers believe there is a 50 per cent chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 years," according to the results published on arXiv. It's a tricky question and opinions vary wildly. Elon Musk and Stephen Hawking have been vocal about humanity's doom at AI takeover, whereas researchers like Andrew Ng, former chief scientist at Baidu, and Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence, have a more conservative attitude. The survey asks the question: "When will AI exceed human performance?"

Editor's note: You may want to check out part 1 of this tutorial before proceeding. In the previous article, we used Tensorflow (TF) to build and learn a linear regression model with a single feature so that given a feature value (house size/sqm), we can predict the outcome (house price/$). In machine learning (ML) literature, we come across the term'training' very often, let us literally look at what that means in TF. The goal in linear regression is to find W, b, such that given any feature value (x), we can find the prediction (y) by substituting W, x, b values into the model. However to find W, b that can give accurate predictions, we need to'train' the model using available data (the multiple pairs of actual feature (x), and actual outcome (y_), note the underscore).