More than three million students across California traded in pencils for computers to take their standardized tests last school year. You might have read about the statewide results of the California Assessment of Student Performance and Progress: More than half of the state's public school students in grades 3 to 8 and 11th grade failed to meet benchmarks for college readiness. The test is new and considered harder than previous ones -- and scores did increase from 2015, the first year scores were reported. But they remained low -- and certain groups, such as black students, lagged behind. Some schools and districts performed very well.

Artificial intelligence (AI), deep learning, and neural networks represent incredibly exciting and powerful machine learning-based techniques used to solve many real-world problems. For a primer on machine learning, you may want to read this five-part series that I wrote. While human-like deductive reasoning, inference, and decision-making by a computer is still a long time away, there have been remarkable gains in the application of AI techniques and associated algorithms. The concepts discussed here are extremely technical, complex, and based on mathematics, statistics, probability theory, physics, signal processing, machine learning, computer science, psychology, linguistics, and neuroscience. That said, this article is not meant to provide such a technical treatment, but rather to explain these concepts at a level that can be understood by most non-practitioners, and can also serve as a reference or review for technical folks as well.

Mario Fritz heads the Scalable Learning and Perception group at the Max Planck Institute for Informatics. Advances in sensor technology and availability of data resources on the Web now give machines a more detailed "picture" of the real world than ever before--but machines are not yet able to acquire the rich semantic understanding that comes easily to humans. To narrow this gap, two of the group's main research themes are scalable learning, to facilitate the acquisition of large-scale knowledge representations, and scalable inference, to enable reasoning over large output spaces at test time. Progress in this direction will facilitate seamless interaction and information exchange between machines and humans, with applications in information retrieval, robotics, human-machine collaboration, and assisted living.

Neota Logic has extended its law school partnership programme, this time with the Faculty of Law at the University of Technology Sydney (UTS) in Australia. The legal AI and expert systems company's most recent outreach venture will see 20 UTS students develop AI applications to improve the delivery of social justice, which in this case centres around working with not-for-profits. The project, that launches next Spring is also supported by Australian law firm, Allens, whose partners will be involved directly in the programme. Allens is also the alliance partner of UK Magic Circle law firm, Linklaters. The latest educational venture follows on from several others, including last October's partnership with Melbourne Law School, where the aim was to build websites to provide legal help to the public. The programme dealt with common legal problems including inaccurate credit reports, handling and managing fines, and assessing employment rights.

Writing in the Times Higher Education, Professor Rose Luckin gives four reasons why higher education needs to embrace the positives of Artificial Intelligence (AI) and not just focus on the negatives. To view this article, you need to register with the Times Higher Education. Please note that registration is free.

Artificial intelligence is very interesting topic of research for many modern scientists. The concept of machine intelligence is really fascinating. It gives human a power to design something that can live on its own. The AI technology has become really advanced and its only matter of time when the machines will be able to learn almost anything. The machine learning algorithms are already very smart, however the processing power has been a challenge in last decade.

Free catered lunch and a dog-friendly office are two of the perks offered by an educational technology company in Palo Alto, Calif., that's looking to hire a machine learning engineer. The position, posted on Dice, will pay between 140,000 and 160,000 to the right candidate who's skilled in machine learning platforms as well as data mining, statistical modeling, and natural language processing. Job-seekers who possess those skills typically could expect multiple job offers, says Matt Leighton, director of recruitment at Mondo, which specializes in digital marketing and technology staffing. The job titles vary from company to company; some might post positions in search of a data scientist or machine learning engineer, others might be after a natural language processing (NLP) programmer or cognitive computing engineer. But hiring companies are seeking the same talent: "They're people who create algorithms through code that allow computers to self-learn," Leighton says.

Although a great methodological effort has been invested in proposing competitive solutions to the class-imbalance problem, little effort has been made in pursuing a theoretical understanding of this matter. In order to shed some light on this topic, we perform, through a novel framework, an exhaustive analysis of the adequateness of the most commonly used performance scores to assess this complex scenario. We conclude that using unweighted H\"older means with exponent $p \leq 1$ to average the recalls of all the classes produces adequate scores which are capable of determining whether a classifier is competitive. Then, we review the major solutions presented in the class-imbalance literature. Since any learning task can be defined as an optimisation problem where a loss function, usually connected to a particular score, is minimised, our goal, here, is to find whether the learning tasks found in the literature are also oriented to maximise the previously detected adequate scores. We conclude that they usually maximise the unweighted H\"older mean with $p = 1$ (a-mean). Finally, we provide bounds on the values of the studied performance scores which guarantee a classifier with a higher recall than the random classifier in each and every class.

We revisit the elegant observation of T. Cover '65 which, perhaps, is not as well-known to the broader community as it should be. The first goal of the tutorial is to explain---through the prism of this elementary result---how to solve certain sequence prediction problems by modeling sets of solutions rather than the unknown data-generating mechanism. We extend Cover's observation in several directions and focus on computational aspects of the proposed algorithms. The applicability of the methods is illustrated on several examples, including node classification in a network. The second aim of this tutorial is to demonstrate the following phenomenon: it is possible to predict as well as a combinatorial "benchmark" for which we have a certain multiplicative approximation algorithm, even if the exact computation of the benchmark given all the data is NP-hard. The proposed prediction methods, therefore, circumvent some of the computational difficulties associated with finding the best model given the data. These difficulties arise rather quickly when one attempts to develop a probabilistic model for graph-based or other problems with a combinatorial structure.

Here's How Google Will Use A.I. to Help Fight Cancer Could killer AI robots bring down America? How UC Berkeley's New Center Could Prevent a Military A.I. Apocalypse Beauty.AI App the 1st international beauty contest judged by AI A treasure hunter went missing in the Rocky Mountains, and a computer algorithm found him ... Drive.ai wants to give self-driving cars more brainpower, personality