With this positional paper we present a representation learning view on predicate invention. The intention of this proposal is to bridge the relational and deep learning communities on the problem of predicate invention. We propose a theory reconstruction approach, a formalism that extends autoencoder approach to representation learning to the relational settings. Our intention is to start a discussion to define a unifying framework for predicate invention and theory revision.

This paper addresses the problem of blind demixing of instantaneous mixtures in a multiple-input multiple-output communication system. The main objective is to present efficient blind source separation (BSS) algorithms dedicated to moderate or high-order QAM constellations. Four new iterative batch BSS algorithms are presented dealing with the multimodulus (MM) and alphabet matched (AM) criteria. For the optimization of these cost functions, iterative methods of Givens and hyperbolic rotations are used. A pre-whitening operation is also utilized to reduce the complexity of design problem. It is noticed that the designed algorithms using Givens rotations gives satisfactory performance only for large number of samples. However, for small number of samples, the algorithms designed by combining both Givens and hyperbolic rotations compensate for the ill-whitening that occurs in this case and thus improves the performance. Two algorithms dealing with the MM criterion are presented for moderate order QAM signals such as 16-QAM. The other two dealing with the AM criterion are presented for high-order QAM signals. These methods are finally compared with the state of art batch BSS algorithms in terms of signal-to-interference and noise ratio, symbol error rate and convergence rate. Simulation results show that the proposed methods outperform the contemporary batch BSS algorithms.

This paper describes a new method for classifying a dataset that partitions elements into their categories. It has relations with neural networks but a slightly different structure, requiring only a single pass through the classifier to generate the weight sets. A grid-like structure is required as part of a novel idea of converting a 1-D row of real values into a 2-D structure of value bands. Each cell in any band then stores a distinct set of weights, to represent its own importance and its relation to each output category. During classification, all of the output weight lists can be retrieved and summed to produce a probability for what the correct output category is. The bands possibly work like hidden layers of neurons, but they are variable specific, making the process orthogonal. The construction process can be a single update process without iterations, making it potentially much faster. It can also be compared with k-NN and may be practical for partial or competitive updating.

Microsoft CEO Satya Nadella, shown here at the company's shareholder meeting in December 2015. Satya Nadella is a believer in the vast promise of artificial intelligence. But the Microsoft CEO says humans and machines need to work together to solve the world's great societal challenges, including issues of diversity and inequality. And responsibility is in the hands of the designers. "Ultimately, it's not going to be about human vs. machine," Nadella wrote in a piece on Slate later reposted to LinkedIn, the professional networking service Microsoft is buying.

It was at Carnegie Mellon where Dr. Hon began seriously building the foundation for his later work in machine-human interaction. His PhD supervisor, Turing Award winner Raj Reddy, was a former student of John McCarthy, the computer scientist who coined the term'artificial intelligence' in 1956 and is widely known as the father of AI. This connection would have richly benefited Dr. Hon, except that an'AI Winter' was happening from 1986 to 1992 when he was completing his doctorate degree. Recalling the climate of this period, "government agencies, universities and even companies slowed down or stopped funding to the field. It was only until the first decade of the 2000's that AI got so hot," said Dr. Hon. Dr. Hon credits the power of improved hardware and software and Big Data for heating up the AI scene.

In the field of artificial intelligence, the phrase deep learning applies to software that improves its model of reality with experience. Consider, for example, a project developed at Google in 2012, in which a neural network running on 16,000 computer processors, browsing through 10 million YouTube videos, began on its own to identify and seek out one of the most popular YouTube genres: cat videos. The then director of that project, Andrew Ng, went on to become the founding chief scientist at Baidu Research, an innovation center run by the giant Web services company Baidu. The parent company owns the largest search engine in China, along with Chinese-language browsers, online encyclopedias, social networks, and other Web-based services. According to the company, Baidu responds to more than 6 billion search requests from more than 138 countries every day. Because search engines and advertising placement platforms (such as Baidu's Phoenix Nest) depend on artificial intelligence (AI) to satisfy vague or ambiguous requests, the company -- along with Google, Microsoft, and other providers of internet guidance -- has a natural interest in machine learning.

Daniel Schwartzkopff, commercial director and DataProphet co-founder, says the start-up is looking towards the European and North American markets for further expansion. Yellowwoods Capital Holdings, a member of the European-based global investment and private equity focused Yellowwoods Group, has acquired a significant interest in DataProphet, where the local start-up will act as the advanced analytics partner for the group. The group's local investments include Hollard Insurance, Clientele and Nandos, among others. While the actual value of the investment into the start-up remains under wraps, DataProphet commercial director and co-founder, Daniel Schwartzkopff, says the business has hit multimillion-dollar status. "As a private fund, our investor partners would prefer us not to discuss the amount. It can be disclosed, however, that DataProphet was priced at a multimillion-dollar valuation," he explains.

After fifty years of soaring crop yields thanks to fertilizers, pest control, and irrigation, that growth is bottoming out. We solved the food shortfall in the 20th century, but we need to do it again in this century. The UN says crop production must rise 70% by 2050 to meet demand. Startups see cheap sensors and artificial intelligence as the solution. Clever algorithms are processing a deluge of high-resolution data enabling real-time monitoring of crops and their environment for the first time.

And at this juncture in the Internet's evolution it seems very plain, Dear Online Reader, that you are mostly being served a tsunami of content -- accelerated into your attention trough by click-dependent digital business models which require a steady stream of word fodder to engage eyeballs long enough to ambush them with ads. You can weep for the state of the written word -- and let's face it, a whole lot of content isn't even written these days; I mean why expend energy composing actual sentences when you can just livestream a melon being slow-ploded by rubber bands? The second option is what Israeli startup Articoolo has been busy doing. It's built an algorithm that can generate an article on a topic of your choice -- so long as it can be described in between two and five words. Just sum up your subject concisely, tell the machine how many words (max 500) to scribe on your behalf, and select whether you prefer "better readability" or "enhanced uniqueness".

Physicist Stephen Hawking says pollution, greed and stupidity are the greatest threats to Earth. Hawking told Larry King Now on Saturday that he's worried by overcrowding. "We certainly have not become less greedy or less stupid," Hawking said. "Six years ago I was worrying about pollution and overcrowding. They have gotten worse since then."