This work posits a way to integrate first order logic rules with neural networks structures. It enables to cooperate expert knowledge with the workhorse deep neural networks. For being more specific, given a sentiment analysis problem, you know that if there is "but" in the sentence the sentiment content changes direction along the sentence. Such rules are harnessed with the network. The method combines two precursor ideas of information distilling [Hinton et al. 2015] and posterior regularization [Ganchev et al. 2010].

MERL researchers have unveiled "Deep Psychic", a futuristic machine learning method that takes pattern recognition to the next level, by not only recognizing patterns, but also predicting them in the first place. The technology uses a novel type of time-reversed deep neural network called Loopy Supra-Temporal Meandering (LSTM) network. The network was trained on multiple databases of historical expert predictions, including weather forecasts, the Farmer's almanac, the New York Post's horoscope column, and the Cambridge Fortune Cookie Corpus, all of which were ranked for their predictive power by a team of quantitative analysts. The system soon achieved super-human performance on a variety of baselines, including the Boca Raton 21 Questions task, Rorschach projective personality test, and a mock Tarot card reading task. Deep Psychic has already beat the European Psychic Champion in a secret match last October when it accurately predicted: "The harder the conflict, the more glorious the triumph."

After a drawn-out battle, South Korea's Go grandmaster with 9-dan rank, Lee Sedol, lost his fifth game against Google's artificial intelligence (AI) program AlphaGo in Seoul on March 15, 2016. AlphaGo's win over one of the world's best players shocked the world's Go circle. Due to the complexity of the nature of Go, which requires intuition, creativity, and strategic thinking, it was believed that Go was the only board game that no computers could conquer. Hong Kong's Go champion, Lee Cheuk-leung, was surprised at the result of the fifth match, in which Lee Sedol had the upper hand in the first half of the game, but somehow lost to the computer eventually. Experts from the Go circle initially expected Lee Sedol to win all five games, but he ultimately lost four of them to the computer.

Robot workers ready to take your call? No one expected a computer to beat a human at the ancient Korean board game Go for another few years at least. So when Google's "AlphaGo" artificial intelligence won against champion player Lee Sedol last month, there were ripples of shock and awe. A far more complex game than chess, this was a "holy grail" moment for machine learning, an important milestone in history. South Korean professional Go player Lee Sedol, right, puts the first stone against Google's artificial intelligence program, AlphaGo, during the Google DeepMind Challenge Match in Seoul.

The sheer volume and unstructured nature of the data generated by billions of connected devices and systems presents significant challenges for those in search of turning this data into insight. For many, machine learning holds the promise of not only structuring this vast amount of data but also to create true business intelligence that can be monetized and leveraged to guide decisions. In the past, it wasn't possible or practical to implement machine learning at such a large scale for a variety of reasons. Machine learning, generally speaking, refers to a class of algorithms that learn from data, uncover insights, and predict behavior without being explicitly programmed. Machine learning algorithms vary greatly depending on the goal of the enterprise and can include various algorithms targeting classification or anomaly detection, clustering of information, time series prediction such as video and speech and even state-action learning and decision making through the use of reinforcement learning.

Artificial intelligence may seem like a futuristic concept, but we're already experiencing it in real ways in our lives, whether we know it or not -- in areas including speech recognition, spam filters and even loan processing. And AI is only going to get more sophisticated from here. That was one of the messages from Oren Etzioni, CEO of the Seattle-based Allen Institute for Artificial Intelligence (AI2), founded by Microsoft co-founder Paul Allen. Etzioni spoke with us for this week's episode of the GeekWire radio show and podcast. Our conversation comes amid a boom in everyday AI, from self-driving cars to a computer that has mastered the game of Go. Microsoft put its stake in the ground with an AI-driven vision that CEO Satya Nadella calls "Conversation as a Platform," with virtual agents working on our behalf. Etzioni takes a much more optimistic view of AI than some of his peers. "The existential risk is just way overblown," he says. "It's much more likely that an asteroid will strike the Earth and annihilate life as we know it than AI will turn evil. Listen to the show below, download the MP3 here, and continue reading for an edited transcript of this week's show. Todd Bishop: Oren, in your current position, you really have a sense for the state of artificial intelligence. I think a lot of people out there see it in their daily lives in a very primitive form. They're watching Google's DeepMind beat a world champion Go player. The potential of artificial intelligence is there in a rudimentary form. Where are we now today in terms of the state of artificial intelligence, and where do you think we'll go over the next three to five years? Oren Etzioni: I do actually think that people are using it more than they realize. In addition to something like Siri, Google Search algorithm uses AI and machine learning all the time. Speech dictation on our phones whether it's Android or iPhone has gotten tremendously better and that's using deep learning behind the scenes to improve what's called a speech recognition. Loan processing these days is often done in a highly automated fashion using machine learning. As a matter of fact, AI is becoming more invisible and integrated into our lives. Of course, that can be a little bit scary to people. They say, "Wait a minute.

Go looks simple, deceptively so. The Chinese board game is played on a board with a grid of 19x19 lines. The object is for two players to alternately place black and white markers on vacant intersections of those lines. And now, this nearly 3,000-year-old board game is a frontier of Artificial Intelligence development. At the time of writing, Google's DeepMind AI's AlphaGo program has played four games of a five game series against Go world champion, South Korea's Lee se-Dol.

Computers that draw and analyze data from the Internet are ubiquitous in many industries. But the new wave of deep-learning machines makes this technology far more compelling for military use. This is attributed to the proliferation of data collectors like drones and smart devices -- known as the "Internet of things" -- combined with advances in software algorithms and the vast computing power available in the cloud. How the Pentagon could use smart machines to gain an edge on the battlefield is now the subject of many closed-door conversations and exchanges with the intelligence community and Silicon Valley firms.

There is huge potential for deep machine learning to become a valuable asset in the intelligence gathering space, according to Pentagon Deputy Secretary Robert Work -- it could ultimately allow U.S. forces to get an edge in the fight against the Islamic State of Iraq and Syria (ISIS, ISIL, IS) by providing greater insights into their networks and practices. The evaluative capabilities and intelligence gathering promise of deep machine learning, Work said, has already shown great potential through the use of publicly available materials on social media, which paint a clearer picture of the events surrounding the downing of Malaysian passenger airliner MH17. Growing tensions between Russia and China were also discussed as a point of concern around the potential for machines to be given lethal authority and how the U.S. might respond in such a case. "There are two things that really keep me up at night about this competition; the first is adversaries who will give machines lethal authority and how will we respond to that," he said.

There is huge potential for deep machine learning to become a valuable asset in the intelligence gathering space, according to Pentagon Deputy Secretary Robert Work -- it could ultimately allow U.S. forces to get an edge in the fight against the Islamic State of Iraq and Syria (ISIS, ISIL, IS) by providing greater insights into their networks and practices. Work made the statement during a roughly hour-long talk called Securing Tomorrow, held March 30 by the Washington Post, where he addressed some of the threat concerns facing the United States and the strategy the Department of Defense is deploying to overcome them. Moderated by Post columnist David Ignatius, the discussion also focused on how the behemoth agency is approaching new technologies and the perceived threats being seen from top international competitors, like Russia and China. "Without question, we are absolutely certain, that the use of deep learning machines is going to allow us to have a better understanding of ISIS as a network and a better understanding of how we can target it precisely and lead to its defeat." The evaluative capabilities and intelligence gathering promise of deep machine learning, Work said, has already shown great potential through the use of publicly available materials on social media, which paint a clearer picture of the events surrounding the downing of Malaysian passenger airliner MH17.