With a paralyzing spinal cord injury, the biological wiring that hooks up our controlling brains to our useful limbs gets snipped, leading to permanent loss of sensation and control and usually a lifetime of extra health care. Researchers have spent years working to repair those lost connections, allowing paralyzed patients to sip coffee and enjoy a beer with robotic limbs controlled by just their minds. Now, researchers have gone a step further, allowing a paralyzed person to control his own hand with just his mind. In a study published Wednesday in Nature, researchers report using a "neural bypass" that reconnects a patient's mental commands for movement to responsive muscles in his limbs, creating somewhat of a synthetic nervous system. The pioneering patient, Ian Burkhart, a 24-year-old man left with quadriplegia after a diving accident almost six years ago, can once again move his hand.

More than 70 years ago, Isaac Asimov dreamed up his three laws of robotics, which insisted, above all, that "a robot may not injure a human being or, through inaction, allow a human being to come to harm". Now, after Stephen Hawking warned that "the development of full artificial intelligence could spell the end of the human race", two academics have come up with a way of teaching ethics to computers: telling them stories. Mark Riedl and Brent Harrison from the School of Interactive Computing at the Georgia Institute of Technology have just unveiled Quixote, a prototype system that is able to learn social conventions from simple stories. Or, as they put in their paper Using Stories to Teach Human Values to Artificial Agents, revealed at the AAAI-16 Conference in Phoenix, Arizona this week, the stories are used "to generate a value-aligned reward signal for reinforcement learning agents that prevents psychotic-appearing behaviour". A simple version of a story could be about going to get prescription medicine from a chemist, laying out what a human would typically do and encounter in this situation.

Time to get away from it all, enjoy our families, friends, and free time... and read up on the latest in data science, machine learning, and analytics. For those of us who can't completely disconnect, or are otherwise interested in reading up over the weekend, the following is a roundup of some of the best free recent ebooks and other online reading resources, as well as a classic throwback article worthy of the attention of newcomers to the field of machine learning. As reported earlier this week, the MIT Press Deep Learning book is finished, and the online version has been finalized. Written by deep learning heavyweights Ian Goodfellow, Yoshua Bengio, and Aaron Courville, the book is poised to become the deep learning book on the market. At over 700 pages, and being quite technical in content, this isn't a simple one-weekend read (at least, not for the majority of folks), but getting started this weekend means only a few more needed.

Welcome to the seventh part of our machine learning regression tutorial within our Machine Learning with Python tutorial series. Up to this point, you have been shown the value of linear regression and how to apply it with Scikit Learn and Python, now we're going to dive into how it is calculated. While I do not believe it is necessary to dig into all of the math that goes into every machine learning algorithm (have you dug into the source code of your other favorite modules to see how they do every little thing?), linear algebra is essential to machine learning, and it is useful to understand the true building blocks that machine learning is built upon. The objective of linear algebra is to calculate relationships of points in vector space. This is used for a variety of things, but one day, someone got the wild idea to do this with features of a dataset.

People have an upper hand over artificial intelligence (AI) when it concerns intuitive thinking and solving complex science problems, according to a new study. In the past few decades, the progress in science and technology have enabled scientists to develop AI that beats people at their own games, however the new discovery reveals a different angle. Associate Professor Jacob Sherson from the Aarhus University (AU) in Denmark led a team of researchers to create a quantum computing game based around complex theoretical science. Later on, it was found that computerized numerical optimization failed to find solutions for the tough problems associated with quantum computing tasks, whereas the human players were successful at it. "The big surprise we had was that some of the players actually had solutions that were of higher quality and of shorter duration than any computer algorithms could find," Jacob Sherson said.

Kaggle is a Data Science community where thousands of Data Scientists compete to solve complex data problems. In this article we are going to see how to go through a Kaggle competition step by step. The contest explored here is the San Francisco Crime Classification contest. The goal is to classify a crime occurrence knowing the time and place it happened. Here, the objectives are fixed by Kaggle.

This book covers the topic of recommender systems comprehensively, starting with the fundamentals and then exploring the advanced topics. Algorithms and evaluation: These chapters discuss the fundamental algorithms in recommender systems, including collaborative filtering methods, content-based methods, knowledge-based methods, ensemble-based methods, and evaluation. Recommendations in specific domains and contexts: The context of a recommendation can be viewed as important side information that affects the recommendation goals. Different types of context such as temporal data, spatial data, social data, tagging data, and trustworthiness are explored. Advanced topics and applications: Various robustness aspects of recommender systems, such as shilling systems, attack models, and their defenses are discussed.

Time-varying mixture densities occur in many scenarios, for example, the distributions of keywords that appear in publications may evolve from year to year, video frame features associated with multiple targets may evolve in a sequence. Any models that realistically cater to this phenomenon must exhibit two important properties: the underlying mixture densities must have an unknown number of mixtures, and there must be some "smoothness" constraints in place for the adjacent mixture densities. The traditional Hierarchical Dirichlet Process (HDP) may be suited to the first property, but certainly not the second. This is due to how each random measure in the lower hierarchies is sampled independent of each other and hence does not facilitate any temporal correlations. To overcome such shortcomings, we proposed a new Smoothed Hierarchical Dirichlet Process (sHDP). The key novelty of this model is that we place a temporal constraint amongst the nearby discrete measures $\{G_j\}$ in the form of symmetric Kullback-Leibler (KL) Divergence with a fixed bound $B$. Although the constraint we place only involves a single scalar value, it nonetheless allows for flexibility in the corresponding successive measures. Remarkably, it also led us to infer the model within the stick-breaking process where the traditional Beta distribution used in stick-breaking is now replaced by a new constraint calculated from $B$. We present the inference algorithm and elaborate on its solutions. Our experiment using NIPS keywords has shown the desirable effect of the model.

Humans are able to accelerate their learning by selecting training materials that are the most informative and at the appropriate level of difficulty. We propose a framework for distributing deep learning in which one set of workers search for the most informative examples in parallel while a single worker updates the model on examples selected by importance sampling. This leads the model to update using an unbiased estimate of the gradient which also has minimum variance when the sampling proposal is proportional to the L2-norm of the gradient. We show experimentally that this method reduces gradient variance even in a context where the cost of synchronization across machines cannot be ignored, and where the factors for importance sampling are not updated instantly across the training set.

This is the latest installment in a regular series of conversations with William McDonough (@billmcdonough), designer, architect, author and entrepreneur. Joel Makower: Tell me about the innovation future roundtable you recently convened. Bill McDonough: I have been working with companies that are looking at the future of mobility in India, and designing factories and other things for them. The chairman said he would like to connect to some of the advanced thinking across many sectors and integrate that with some conversations that he could participate in. The first person I thought of for that was Jack Hidary.