High throughput screening of compounds (chemicals) is an essential part of drug discovery [7], involving thousands to millions of compounds, with the purpose of identifying candidate hits. Most statistical tools, including the industry standard B-score method, work on individual compound plates and do not exploit cross-plate correlation or statistical strength among plates. We present a new statistical framework for high throughput screening of compounds based on Bayesian nonparametric modeling. The proposed approach is able to identify candidate hits from multiple plates simultaneously, sharing statistical strength among plates and providing more robust estimates of compound activity. It can flexibly accommodate arbitrary distributions of compound activities and is applicable to any plate geometry. The algorithm provides a principled statistical approach for hit identification and false discovery rate control. Experiments demonstrate significant improvements in hit identification sensitivity and specificity over the B-score method, which is highly sensitive to threshold choice. The framework is implemented as an efficient R extension package BHTSpack and is suitable for large scale data sets.

In phase retrieval we want to recover an unknown signal $\boldsymbol x\in\mathbb C^d$ from $n$ quadratic measurements of the form $y_i = |\langle{\boldsymbol a}_i,{\boldsymbol x}\rangle|^2+w_i$ where $\boldsymbol a_i\in \mathbb C^d$ are known sensing vectors and $w_i$ is measurement noise. We ask the following weak recovery question: what is the minimum number of measurements $n$ needed to produce an estimator $\hat{\boldsymbol x}(\boldsymbol y)$ that is positively correlated with the signal $\boldsymbol x$? We consider the case of Gaussian vectors $\boldsymbol a_i$. We prove that - in the high-dimensional limit - a sharp phase transition takes place, and we locate the threshold in the regime of vanishingly small noise. For $n\le d-o(d)$ no estimator can do significantly better than random and achieve a strictly positive correlation. For $n\ge d+o(d)$ a simple spectral estimator achieves a positive correlation. Surprisingly, numerical simulations with the same spectral estimator demonstrate promising performance with realistic sensing matrices. Spectral methods are used to initialize non-convex optimization algorithms in phase retrieval, and our approach can boost the performance in this setting as well. Our impossibility result is based on classical information-theory arguments. The spectral algorithm computes the leading eigenvector of a weighted empirical covariance matrix. We obtain a sharp characterization of the spectral properties of this random matrix using tools from free probability and generalizing a recent result by Lu and Li. Both the upper and lower bound generalize beyond phase retrieval to measurements $y_i$ produced according to a generalized linear model. As a byproduct of our analysis, we compare the threshold of the proposed spectral method with that of a message passing algorithm.

A challenge in training discriminative models like neural networks is obtaining enough labeled training data. Recent approaches use generative models to combine weak supervision sources, like user-defined heuristics or knowledge bases, to label training data. Prior work has explored learning accuracies for these sources even without ground truth labels, but they assume that a single accuracy parameter is sufficient to model the behavior of these sources over the entire training set. In particular, they fail to model latent subsets in the training data in which the supervision sources perform differently than on average. We present Socratic learning, a paradigm that uses feedback from a corresponding discriminative model to automatically identify these subsets and augments the structure of the generative model accordingly. Experimentally, we show that without any ground truth labels, the augmented generative model reduces error by up to 56.06% for a relation extraction task compared to a state-of-the-art weak supervision technique that utilizes generative models.

There's a reason you've never seen fully autonomous exoskeletons that help the disabled walk without crutches: Building one is crazy hard. But the founders of a Paris-based startup called Wandercraft are uniquely qualified to do it. They're roboticists who happen to have loved ones in wheelchairs, giving them both the expertise and motivation to develop an exoskeleton that helps users walk again. After years of development, they're nearly ready to show it to the public, following a round of promising patient trials. Wandercraft ran successful preliminary trials with a handful of clients using "Atalante," its latest prototype.

Many people in Silicon Valley believe in the Singularity--the day in our near future when computers will surpass humans in intelligence and kick off a feedback loop of unfathomable change. When that day comes, Anthony Levandowski will be firmly on the side of the machines. In September 2015, the multi-millionaire engineer at the heart of the patent and trade secrets lawsuit between Uber and Waymo, Google's self-driving car company, founded a religious organization called Way of the Future. Its purpose, according to previously unreported state filings, is nothing less than to "develop and promote the realization of a Godhead based on Artificial Intelligence." Mark Harris is a freelance journalist reporting on technology from Seattle. Sign up to get Backchannel's weekly newsletter, and follow us on Facebook and Twitter. Way of the Future has not yet responded to requests for the forms it must submit annually to the Internal Revenue Service (and make publically available), as a non-profit religious corporation. However, documents filed with California show that Levandowski is Way of the Future's CEO and President, and that it aims "through understanding and worship of the Godhead, [to] contribute to the betterment of society." A divine AI may still be far off, but Levandowski has made a start at providing AI with an earthly incarnation. The autonomous cars he was instrumental in developing at Google are already ferrying real passengers around Phoenix, Arizona, while self-driving trucks he built at Otto are now part of Uber's plan to make freight transport safer and more efficient. He even oversaw a passenger-carrying drones project that evolved into Larry Page's Kitty Hawk startup. Levandowski has done perhaps more than anyone else to propel transportation toward its own Singularity, a time when automated cars, trucks and aircraft either free us from the danger and drudgery of human operation--or decimate mass transit, encourage urban sprawl, and enable deadly bugs and hacks. But before any of that can happen, Levandowski must face his own day of reckoning.

This is an era of data-centric computing. For those in hardware engineering who embraced disruption over the past few years – my colleagues at IBM Cognitive Systems and our partner NVIDIA included – this sudden rise of AI-inspired applications is a once or twice-in-a-career thrill. From a practical perspective, we quickly realize that unbounded bandwidth, a sort of information superhighway inside computers with no speed limit, is a condition of unbounded software and algorithm innovation. As my generation learned so painfully, trying to use digital audio and video media in the early days of the internet, the best computer or network is the one that you don't have to think about or wait on endlessly. We're excited about the launch of NVIDIA's Volta GPU accelerators.Together with the NVIDIA NVLink "information superhighway" at the core of our IBM Power Systems, it provides what we believe to be the closest thing to an unbounded platform for those working in machine learning and deep learning and those dealing with very large data sets.

The cybersecurity industry has always had a fortress mentality: Firewall the perimeter! But that mindset has failed--miserably, as each new headline-generating hack reminds us. Even if you do patch all your software, the way Equifax didn't, or you randomize all your passwords, the way most of us don't, bad actors are going to get past your heavily guarded gate, into your network. And once they do, they're free to go wild. Scott Rosenberg is an editor at Backchannel.

FILE - This Sept. 6, 2012, file photo shows the Amazon logo in Santa Monica, Calif. Amazon unveils new devices, Wednesday, Sept. 27, 2017. SEATTLE (AP) -- Amazon says it is cutting the price of its Echo smart speaker to $100 from $180,, improving the sound quality and upgrading its appearance with six new "shells." The next generation speaker, which is powered by Amazon's Alexa voice assistant, will have a dedicated woofer and a tweeter for the first time, as well as Dolby sound. The company made the announcement in Seattle at an event for journalists.

Laser weapons are on a roll. The U.S. Air Force, Army, Navy, Marines, and the Joint Improvised-Threat Defeat Organization are testing them. Plans include mounting them on Humvees to shoot down drones. You can see them destroy drones on YouTube. The Missile Defense Agency wants to test laser-equipped drones as a defense against North Korean missiles.

The UN has warned that robots could destabilise the world ahead of the opening of a headquarters in The Hague to monitor developments in artificial intelligence. From the risk of mass unemployment to the deployment of autonomous robotics by criminal organisations or rogue states, the new Centre for Artificial Intelligence and Robotics has been set the goal of second-guessing the possible threats. It is estimated that 30% of jobs in Britain are potentially under threat from breakthroughs in artificial intelligence, according to the consultancy firm PwC. In some sectors half the jobs could go. A recent study by the International Bar Association claimed robotics could force governments to legislate for quotas of human workers.