We study large-scale kernel methods for acoustic modeling in speech recognition and compare their performance to deep neural networks (DNNs). We perform experiments on four speech recognition datasets, including the TIMIT and Broadcast News benchmark tasks, and compare these two types of models on frame-level performance metrics (accuracy, cross-entropy), as well as on recognition metrics (word/character error rate). In order to scale kernel methods to these large datasets, we use the random Fourier feature method of Rahimi and Recht (2007). We propose two novel techniques for improving the performance of kernel acoustic models. First, in order to reduce the number of random features required by kernel models, we propose a simple but effective method for feature selection. The method is able to explore a large number of non-linear features while maintaining a compact model more efficiently than existing approaches. Second, we present a number of frame-level metrics which correlate very strongly with recognition performance when computed on the heldout set; we take advantage of these correlations by monitoring these metrics during training in order to decide when to stop learning. This technique can noticeably improve the recognition performance of both DNN and kernel models, while narrowing the gap between them. Additionally, we show that the linear bottleneck method of Sainath et al. (2013) improves the performance of our kernel models significantly, in addition to speeding up training and making the models more compact. Together, these three methods dramatically improve the performance of kernel acoustic models, making their performance comparable to DNNs on the tasks we explored.

Think about your typical day. Your phone alarm wakes you up (maybe after hitting snooze a couple of times), you check the weather as your coffee is brewing, and you get traffic alerts about your commute to work. How many times do you think artificial intelligence (AI) is a part of your day? It's definitely more often than you think. During a conversation with President Barack Obama, technologist Joi Ito, director of the MIT Media Lab, predicted that 2016 would be the "year artificial intelligence becomes more than just a computer science problem." In 2016, AI's fast progress led to acquisition deals topping $1 billion, and the White House released its first report on artificial intelligence and the economy.

When a citizen dials 311, it has been the longstanding preference by mayors to have a city employee on the other end of the line to deliver the ever-valuable personal touch. But when efficiency is the priority, are we really best served by having city employees at 311 call centers act primarily as switchboard operators, sifting through online scripts or, worse, binders or spreadsheets and responding to information requests with specifically coded responses? More often than not, citizens will call 311 with an information request -- to determine their trash pick-up day, the hours the public pool is open or another simple ask. The North Carolina Innovation Center is now using chatbots for its internal IT help desk hotline, where between 80 and 90 percent of calls are for help changing a password. Still in its experimental stage, North Carolina uses the bots to free up help center operators to handle more challenging and complex concerns.

Businesses and consumers alike have been affected by the almost daily threats of data breaches and the impact of these have on a continuous basis. Will such threats enable identity fraud, send a business under or even give adversaries further power to conduct ever more dangerous attacks? Michael Sutton, CISO at Zscaler, has crafted his top five predictions for the year ahead and what this will mean for the threat landscape. "Offensive Offense – Increasingly, motivations for offensive nation state sponsored attacks have gone into a new realm and have been driven primarily as an effort to undermine the credibility of another government or in some cases influence public sentiment. The Director of National Intelligence went so far as to publicly accuse the Russian Government of the attack on the DNC and others have openly speculated that they too were behind the compromise of Hillary Clinton Campaign Chairman John Podesta's inbox. In light of such aggressive and direct medaling in the political affairs of another nation, some in the intelligence community are suggesting that the US should return the favour. This is a troubling notion. If we enter an era where nations are actively conducting offensive cyber attacks with the primary goal of embarrassing their foe by leaking documents online, many innocent victims will be caught in the crossfire. It's one thing to conduct cyber espionage covertly to get a leg up on the competition either from a military or economic perspective, but it is an entirely different situation when private documents are being handed over to Wikileaks. Given current political tensions, the precedent that has already been set and the aggressive tone of the incoming US administration, it's likely that 2017 will see the US and other nations step into this cyber mudslinging contest."

After a decline in 2015, U.S. patenting activity rose to a record high in 2016. The U.S. Patent and Trademark Office granted 304,126 patents last year, up from 298,407 in 2015. The increase was largely driven by West Coast technology companies, including Amazon, Apple, AT&T, Boeing, IBM, Intel, Google, and Microsoft. According to a 24/7 Wall St. analysis of data provided by Information for Industry (IFI) Claims Patent Services, Korean conglomerate Samsung Electronics was granted the most patents in 2016. Including its major subsidiaries, Samsung was awarded more than 8,500 patents, slightly higher than IBM's 8,088 awarded patents.

The U.S. Department of Transportation will launch a federal committee on self-driving cars that includes multiple notable technology executives, according to an announcement on Thursday. Via a statement, the group will regularly meet to discuss various aspects of automated technology and the Department of Transportation's role in setting policy. The committee, which includes leading professionals and experts in their field, will hold its first meeting on January 16th, 2017 to immediately begin work on some of the most pressing and relevant matters facing transportation today, including the development and deployment of automated vehicles, and determining the needs of the Department as it continues with its relevant research, policy, and regulations. The committee will be chaired by Los Angeles mayor Eric Garcetti and General Motors CEO Mary Barra. Other members include executives and staffers from companies such as Amazon, Lyft and Zipcar.

An incredible drone's-eye video that shows an unfortunate flying robot crashing into Seattle's famous Space Needle has just been released. The bizarre crash, which occurred on New Year's Eve, was captured by the drone's on-board camera. After filming a stunning view of downtown Seattle, the drone, which is hovering near the Space Needle's observation deck, suddenly accelerates and slams into the landmark's roof. Q13 Fox reports that the drone crashed about 575 feet above the ground and notes that the Space Needle was not damaged in the collision. It is not clear who was operating the drone at the time of the incident.

With the use of data-centric business models and big data services on the rise, it is becoming increasingly harder to detect threats and data breaches. Cybersecurity experts are finding themselves hard pressed to keep tabs on the reams of data that are being generated by their companies and organizations. Attackers on the other hand are finding it easier to hide their malicious packets in the flood of data that is being exchanged over corporate networks. Gary Vaynerchuk was so impressed with TNW Conference 2016 he paused mid-talk to applaud us. A solution to this dilemma might be found in the use of machine learning, the hot trend that is taking the world by storm and is transforming numerous industries in ways that were previously inconceivable.

"A reasoned, even cournterempathic analysis of moral obligation and likely consequences is a better guide to planning for the future than the gut wrench of empathy." This is the argument presented by Paul Bloom in his latest book, Against Empathy. He argues that a more fair and reliable guide for one's morality is rational compassion; a morality based on facts, cost/benefit analysis and rational thinking. For the most part, I agree with him. I think back to all of the conflicts that I have witnessed or been engaged in, whether with friends, colleagues or family.

The unstoppable rise of robots in our everyday lives requires urgent EU rules including fitting machines with'kill switches', European Parliament members have warned. A draft report passed by an EU committee today also proposes granting legal status to robots, categorising them as'electronic persons'. MEPs passed the resolution urging Brussels to take action on automaton ethics, as the role of robots in everyday life becomes more and more significant. The warning has echoes of science fiction, such as the Terminator franchise of films, where robots take on humans. MEPs have warned that robots need to be fitted with'kill switches' to prevent a Terminator-style uprising against humans Mady Delvaux, a Socialist MEP from Luxembourg, led the campaign and warned that Europe is passively standing by as robots take an increasingly powerful role that will turn even stronger with the emergence of driverless cars.