Now artificial intelligence is set to conquer the world of music. Swiss researchers say they have developed a computer algorithm that can generate new tunes in various musical genres after listening to more than 2,000 songs. And while it isn't quite at the level of Justin Bieber or Adele, future versions of the system may be fine-tuned to create songs that rival those by today's musicians. The deep artificial composer, or DAC, produces original melodies that are'quite agreeable to listen to', according to its developers at EPFL research university in Lausanne, Switzerland The DAC system is trained to'listen' to existing tunes to learn what works. It then teaches itself to predict the pitch and duration of every note following another.

Insurance and technology are inextricably linked. The risks we seek to manage have always been connected to the tools we use. But, as Accenture's 2017 Technology Vision for Insurance points out, this relationship recently reached an inflection point. In the past, humans generally changed themselves to make use of new technologies--we learned to drive, we learned to type, we learned to code. But now modern technologies are sophisticated enough to adapt themselves to us.

According to new IBM research, released Thursday, more than 73% of CEOs predict that cognitive computing, or artificial intelligence (AI), technologies will play a "key role" in their business's future. The press release announcing the findings also noted that more than 50% of these CEOs plan to adopt these technologies by the year 2019. In terms of where AI will prove most impactful, respondents listed information technology, sales, and information security as the top three priorities for the technology. The 6,000 CEOs surveyed also said they expect a 15% return on their AI investments. In regards to IT, cognitive solutions can improve software development and testing, lead to better efficiency and agility, and boost solution design the release said.

Over the past few decades, technology has dramatically increased the productivity of the average worker, and through them, the profitability of the average large business. It's a trend that has payed enormous dividends for those companies that have been able to adapt, but in the past couple of years the trend seems to have slowed, and corporate profits have stagnated. Industry researchers from Accenture Research and Frontier Economics recently forecasted a dramatic return to growth in corporate productivity and profit. Called How A.I. Boosts Industry Profits and Innovation, their report foresees a number of improvements to profitability, thanks to A.I. moving into the workplace. The most dramatic predictions of the bunch have to do with overall growth in the economy.

Note: This is Part 2 of a two-part series on AI. PDF: We made a fancy PDF of this post for printing and offline viewing. We have what may be an extremely difficult problem with an unknown time to solve it, on which quite possibly the entire future of humanity depends. Welcome to Part 2 of the "Wait how is this possibly what I'm reading I don't get why everyone isn't talking about this" series. Part 1 started innocently enough, as we discussed Artificial Narrow Intelligence, or ANI (AI that specializes in one narrow task like coming up with driving routes or playing chess), and how it's all around us in the world today. We then examined why it was such a huge challenge to get from ANI to Artificial General Intelligence, or AGI (AI that's at least as intellectually capable as a human, across the board), and we discussed why the exponential rate of technological advancement we've seen in the past suggests that AGI might not be as far away as it seems. This left us staring at the screen, ...

COVER A conceptual illustration of an artificial neuron evokes a technology that is transforming many fields of science: artificial intelligence (AI). One common form of AI is a neural network, which "learns" as connections between simulated neurons change in response to inputs. Such systems can find meaningful patterns in vast data sets, ranging from genomics to astronomy, and are even beginning to design experiments.

Companies approach recruiting in different ways. Facebook's method, for example, involves acquisitions for human capital and a six-week onboarding boot camp. Of course, that's not the norm: On average, a recent study found it takes 42 days--and a cost-per-hire of $4,129--to fill an open position. That adds up, especially since Bloomberg recently reported that approximately 10,000 members of the baby boomer generation reach retirement age every day. With numbers like that, it's no wonder efficiency-boosting advancements in machine learning are beginning to challenge the HR status quo.

Predicting the click-through rate of an advertisement is a critical component of online advertising platforms. In sponsored search, the click-through rate estimates the probability that a displayed advertisement is clicked by a user after she submits a query to the search engine. Commercial search engines typically rely on machine learning models trained with a large number of features to make such predictions. This is inevitably requires a lot of engineering efforts to define, compute, and select the appropriate features. In this paper, we propose two novel approaches (one working at character level and the other working at word level) that use deep convolutional neural networks to predict the click-through rate of a query-advertisement pair. Specially, the proposed architectures only consider the textual content appearing in a query-advertisement pair as input, and produce as output a click-through rate prediction. By comparing the character-level model with the word-level model, we show that language representation can be learnt from scratch at character level when trained on enough data. Through extensive experiments using billions of query-advertisement pairs of a popular commercial search engine, we demonstrate that both approaches significantly outperform a baseline model built on well-selected text features and a state-of-the-art word2vec-based approach. Finally, by combining the predictions of the deep models introduced in this study with the prediction of the model in production of the same commercial search engine, we significantly improve the accuracy and the calibration of the click-through rate prediction of the production system.

Reinforcement learning tasks are typically specified as Markov decision processes. This formalism has been highly successful, though specifications often couple the dynamics of the environment and the learning objective. This lack of modularity can complicate generalization of the task specification, as well as obfuscate connections between different task settings, such as episodic and continuing. In this work, we introduce the RL task formalism, that provides a unification through simple constructs including a generalization to transition-based discounting. Through a series of examples, we demonstrate the generality and utility of this formalism. Finally, we extend standard learning constructs, including Bellman operators, and extend some seminal theoretical results, including approximation errors bounds. Overall, we provide a well-understood and sound formalism on which to build theoretical results and simplify algorithm use and development.

In this paper, we study the problem of learning a mixture of Gaussians with streaming data: given a stream of $N$ points in $d$ dimensions generated by an unknown mixture of $k$ spherical Gaussians, the goal is to estimate the model parameters using a single pass over the data stream. We analyze a streaming version of the popular Lloyd's heuristic and show that the algorithm estimates all the unknown centers of the component Gaussians accurately if they are sufficiently separated. Assuming each pair of centers are $C\sigma$ distant with $C=\Omega((k\log k)^{1/4}\sigma)$ and where $\sigma^2$ is the maximum variance of any Gaussian component, we show that asymptotically the algorithm estimates the centers optimally (up to constants); our center separation requirement matches the best known result for spherical Gaussians \citep{vempalawang}. For finite samples, we show that a bias term based on the initial estimate decreases at $O(1/{\rm poly}(N))$ rate while variance decreases at nearly optimal rate of $\sigma^2 d/N$. Our analysis requires seeding the algorithm with a good initial estimate of the true cluster centers for which we provide an online PCA based clustering algorithm. Indeed, the asymptotic per-step time complexity of our algorithm is the optimal $d\cdot k$ while space complexity of our algorithm is $O(dk\log k)$. In addition to the bias and variance terms which tend to $0$, the hard-thresholding based updates of streaming Lloyd's algorithm is agnostic to the data distribution and hence incurs an approximation error that cannot be avoided. However, by using a streaming version of the classical (soft-thresholding-based) EM method that exploits the Gaussian distribution explicitly, we show that for a mixture of two Gaussians the true means can be estimated consistently, with estimation error decreasing at nearly optimal rate, and tending to $0$ for $N\rightarrow \infty$.