Which authors of this paper are endorsers? Disable MathJax (What is MathJax?)

NSA Chief Says Without Artificial Intelligence, Cyber'Is A Losing Strategy' by Giuseppe Macri, Artificial intelligence will play a big role in the future of U.S. strategy in cyberspace, according to National Security Agency Director Adm. Michael Rogers, who told Congress Tuesday that relying primarily on human intelligence "is a losing strategy." "[We're] very much interested in artificial intelligence, machine learning, how we can do cyber at scale [and] at speed," Rogers testified before the Senate Armed Services Committee Tuesday. "Because if we're just going to take this largely human capital approach to doing business, that is a losing strategy." "It will be both incredibly resource intensive, and it will be very slow," Rogers added. "That is a big area of focus for us."

From Google's purchase of Deepmind to Apple's purchase of VocalIQ, big deals in UK's intelligence technologies are booming. However, commercial innovations are only part of the story. Join Digital Catapult to meet SMEs, academics and investors working in this space. It's a fantastic opportunity to find out what the UK's leading academic and research organisations and businesses are doing and attending, you'll also gain exclusive insights into market opportunities, challenges and exciting advances.

Bots may already be old news for you if you are an investor or tech professional in Silicon Valley, where from what I hear they have already been a hot topic for at least the last year. The numbers bear this out: according to CB Insights, artificial intelligence start-ups (of which bots are a portion) raised about 1.6 billion in funding in the first half of 2016. If those involved in building and investing in bots are correct, that 1.6 billion will be peanuts compared to the changes that will come. For those of us not in that rarefied world of cutting edge tech and investors looking for the next thing they can preface with the word "disruptive," bots are (IMHO anyway, being in the relative tech backwater of Boston) only just starting to enter the conversation around how organizations employ technology solutions for improving customer experience management (CEM). CEM professionals therefore need to keep abreast of developments in the bots landscape, since this technology provides new ways that companies can identify, sell to, and service their customers.

We are on the cusp of a major breakthrough in how organizations collect, analyze, and act on knowledge. This article is part of an MIT SMR initiative exploring how technology is reshaping the practice of management. Editor's Note: This article is one of a special series of 14 commissioned essays MIT Sloan Management Review is publishing to celebrate the launch of our new Frontiers initiative. Each essay gives the author's response to this question: "Within the next five years, how will technology change the practice of management in a way we have not yet witnessed?" Artificial intelligence is about to transform management from an art into a combination of art and science.

The online education company Udacity is partnering with major companies in the field of autonomous vehicles to launch a nanodegree program for those interested in becoming a self-driving car engineer. "It is the first and only program of its kind where most people with an internet connection--from Detroit to Damascus and from Adelaide to Aleppo--can learn the skills they need to work in one of the most amazing fields of our time," Udacity founder and president Sebastian Thrun wrote in a blog post. The course, which spans three 12-week terms, covers deep learning, computer vision, sensor fusion, localization and controllers. Four major partners have committed to fast-tracking the nanodegree graduates into positions around the world: Mercedes-Benz, Nvidia, Otto (recently acquired by Uber) and Didi Chuxing. Thrun promised that more partners will be added to that list.

I recently read the paper on GANs, and from what I understand, the networks are trained by making them play a minimax game. I was curious if there is research being done in training networks by having them play more sophisticated games; i.e. expectiminimax or a variation of minimax that only requires a single player. I know GANs are fairly effective, but I surprisingly haven't come across a lot of literature exploring more complicated games. Is it because the notoriously training difficulty of GANs scales with the complexity of the game?

The second issue I noticed was that there seemed to be little connection between the network's confidence in its actions and its actual score. I came across another recent paper on something called Double Q Learning, also courtesy of DeepMind, which substantially improved Google's original results. Double Q Learning counters the tendency for Q networks to become overconfident in their predictions. I changed Google's original Deep Q Network to a Double Deep Q Network, and that helped substantially. Finally, the biggest improvement of all came when I was just more patient. Even running on a powerful machine with a Nvidia 980 GPU, the emulator could only go so fast. As a consequence, one million training steps took about an entire day, with quite a bit of variance in the scores along the way.

Your phone doesn't know how you're feeling -- but you may want it to if that capability came with a few fringe benefits. Affectiva makes emotion-detection software, and CEO Rana el Kaliouby was full of ideas today at Disrupt SF as to how it could be deployed, from gifs to Ubers. "We're obsessed with emotional AI, we wake up thinking about it," said Rana el Kaliouby. "I imagine a future where every device has a little emotion chip and can read your emotions, just like it's touchscreen enabled or GPS enabled." If you aren't creeped out by that, you might see a few of the benefits.

Thanks to the modern gaming industry, we can now spend our evenings wandering around photorealistic game worlds, like the post-apocalyptic Boston of Fallout 4 or Grand Theft Auto V's Los Santos, instead of doing things like "seeing people" and "engaging in human interaction of any kind." Games these days are so realistic, in fact, that artificial intelligence researchers are using them to teach computers how to recognize objects in real life. Not only that, but commercial video games could kick artificial intelligence research into high gear by dramatically lessening the time and money required to train AI. "If you go back to the original Doom, the walls all look exactly the same and it's very easy to predict what a wall looks like, given that data," said Mark Schmidt, a computer science professor at the University of British Columbia (UBC). "But if you go into the real world, where every wall looks different, it might not work anymore." Schmidt works with machine learning, a technique that allows computers to "train" on a large set of labelled data--photographs of streets, for example--so that when let loose in the real world, they can recognize, or "predict," what they're looking at.