One of the huge drawbacks of modern technology is that it fills the air around us with radio signals. From your kitchen radio to your LTE-enabled smartphone, all of these devices use radio waves to communicate. Unfortunately, there is only a certain amount of radio frequencies that can be used. DARPA (Defense Advanced Research Projects Agency) is looking for a way around this problem, and wants teams to develop an artificially intelligent system that will control what devices use what radio waves and when. Basically, instead of forcing devices to make use of narrow frequency bands when the spectrum gets congested, DARPA wants devices to negotiate sharing frequencies when they need them.

Where exactly is all the water on Earth's surface? Stand-alone satellite images have their limitations, but using artificial intelligence to examine them can now glean precise levels of water around the world and how they are changing week by week. Palo Alto startup Orbital Insight uses freely available images taken by the U.S. Geological Survey's Landsat 7 and 8 satellites, much like the images you see on Google Maps. The startup feeds the images into a neural network, which pinpoints the exact location and area of surface water. While it's not a new feat to track the Earth's water levels, Orbital Insight is approaching the task with a specially trained neural network that labels water pixel by pixel.

We consider data in the form of pairwise comparisons of n items, with the goal of precisely identifying the top k items for some value of k < n, or alternatively, recovering a ranking of all the items. We analyze the Copeland counting algorithm that ranks the items in order of the number of pairwise comparisons won, and show it has three attractive features: (a) its computational efficiency leads to speed-ups of several orders of magnitude in computation time as compared to prior work; (b) it is robust in that theoretical guarantees impose no conditions on the underlying matrix of pairwise-comparison probabilities, in contrast to some prior work that applies only to the BTL parametric model; and (c) it is an optimal method up to constant factors, meaning that it achieves the information-theoretic limits for recovering the top k-subset. We extend our results to obtain sharp guarantees for approximate recovery under the Hamming distortion metric, and more generally, to any arbitrary error requirement that satisfies a simple and natural monotonicity condition.

We stabilize the activations of Recurrent Neural Networks (RNNs) by penalizing the squared distance between successive hidden states' norms. This penalty term is an effective regularizer for RNNs including LSTMs and IRNNs, improving performance on character-level language modeling and phoneme recognition, and outperforming weight noise and dropout. We achieve competitive performance (18.6\% PER) on the TIMIT phoneme recognition task for RNNs evaluated without beam search or an RNN transducer. With this penalty term, IRNN can achieve similar performance to LSTM on language modeling, although adding the penalty term to the LSTM results in superior performance. Our penalty term also prevents the exponential growth of IRNN's activations outside of their training horizon, allowing them to generalize to much longer sequences.

In a small startup space just down the street from UC Berkeley's campus, robotics pioneer Homayoon Kazerooni is bragging about how no-frills his invention is. "We're trying to make the Honda," he says, "not the sportscar." Kazerooni is showing me his latest robotic exoskeleton that gives paraplegics and people with mobility problems the ability to stand up from their wheelchairs and walk again. He's been building such bionic systems for more than a decade, and back in 2005 he cofounded Ekso Bionics, the current market leader for exoskeletons. So it's no surprise that Kazerooni says the new device from his new company, SuitX, is the most advanced yet.

With Earth Day come and gone the initiative to save our planet is fresh in our minds. And that's not limited to global warming and putting a stop to pollution; poaching remains a big problem when it comes to preserving the world around us. Luckily researchers have been testing out how we can use scientists to help the fight against poaching. According to Science Daily, organizations such as the National Science Foundation (NSF) and the Army Research Office have teamed up to look into how artificial intelligence (AI) could be used to stop poaching and illegal logging. Led by scientists at the University of South Carolina (USC), researchers have found ways to use game theory, or the mathematical theory of conflict and cooperation, to protect parks in a more proactive way.

If this year's selection of electable decisionmakers feels like it leaves something to be desired, then here's a story for you. Imagine a world without politicians, or, really, politics. Policy options are instead plugged into a sophisticated program and 1,000 or 100,000 simulations are made. Likely outcomes of the policy change are analyzed and unintended consequences are made visible. Politics are reduced to an argument over what inputs should take precedence in the model, not over what decision to make, and the arguments have to contend with data, which takes precedence over all.

One of the most frustrating elements of the current healthcare environment is the administrative burden of prior authorizations for medications and procedures. It is a frustration for providers, for patients, and for payers. Is there any way to solve this dilemma? For physicians, an estimated 20 hours per week is spent in prior authorization activities, costing an average of 83,000 in excess annual overhead per physician. Is there an actual benefit for this effort? Most physicians say that payers (commercial, Medicare, Medicaid, and pharmacy benefit managers (PBMs)) use prior authorizations to keep costs down.

Computer scientists from MIT and a machine learning startup, PatternEx, have reportedly developed a new system that can correctly detect 85% of cyberattacks using artificial intelligence merged with input from human experts. At the moment, security systems are closely monitored by humans and programmed to pick up on cyberattacks that only follow very specific rules, as such missing any attacks that do not follow those rules. But, there are also systems autonomously run by computers that practice anomaly detection – i.e. the identification of items, events or observations – that do not conform to an expected pattern or other items in a dataset. This method often leads to false positives, meaning that humans doubt the reliability of the system and are forced to go back and check all the results anyway. To improve this, researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL), in collaboration with PatternEx, have developed the AI2 artificial intelligent platform, which merges three different machine learning methods that enable computers to learn unsupervised.

The U.S. government believes the Internet of things (IoT) has enormous economic potential across all industries. Its machine-to-machine technologies can reduce automobile-related injuries, usher in an era of precise weather forecasting and automate all types of processes. But what impact will IoT have on jobs? Will it create more than it destroys? And what happens to all the data devices generate?