The analysis of distributed (non-Bayesian) learning algorithm gained popularity since the seminal work of Jadbabaie et al. [1]. The ability of non-Bayesian updates to combine distributed optimization and learning algorithms make them especially useful for the design of distributed estimation algorithms with provable performance. In the distributed learning setup, a group of agents repeatedly receive signals about a certain unknown state of the world or parameter. No single agent has enough information to accurately estimate the unknown state and, thus, interaction with other agents is needed. Several results are readily available for performance evaluation of distributed learning algorithms for a variety of scenarios.

We introduce a novel class of localized atomic environment representations, based upon the Coulomb matrix. By combining these functions with the Gaussian approximation potential approach, we present LC-GAP, a new system for generating atomic potentials through machine learning (ML). Tests on the QM7, QM7b and GDB9 biomolecular datasets demonstrate that potentials created with LC-GAP can successfully predict atomization energies for molecules larger than those used for training to chemical accuracy, and can (in the case of QM7b) also be used to predict a range of other atomic properties with accuracy in line with the recent literature. As the best-performing representation has only linear dimensionality in the number of atoms in a local atomic environment, this represents an improvement both in prediction accuracy and computational cost when considered against similar Coulomb matrix-based methods.

We introduce several techniques for sampling and visualizing the latent spaces of generative models. Replacing linear interpolation with spherical linear interpolation prevents diverging from a model's prior distribution and produces sharper samples. J-Diagrams and MINE grids are introduced as visualizations of manifolds created by analogies and nearest neighbors. We demonstrate two new techniques for deriving attribute vectors: bias-corrected vectors with data replication and synthetic vectors with data augmentation. Binary classification using attribute vectors is presented as a technique supporting quantitative analysis of the latent space. Most techniques are intended to be independent of model type and examples are shown on both Variational Autoencoders and Generative Adversarial Networks.

School of Physical Sciences, University of Chinese Academy of Sciences, Beijing 100190, China Recommender systems play an essential role in the modern business world. They recommend favorable items like books, movies, and search queries to users based on their past preferences. Applying similar ideas and techniques to Monte Carlo simulations of physical systems boosts their e fficiency without sacrificing accuracy. Exploiting the quantum to classical mapping inherent in the continuous-time quantum Monte Carlo methods, we construct a classical molecular gas model to reproduce the quantum distributions. We then utilize powerful molecular simulation techniques to propose e fficient quantum Monte Carlo updates. The recommender engine approach provides a general way to speed up the quantum impurity solvers. At the heart of every quantum Monte Carlo (QMC) method is a quantum to classical mapping.

Experiments in particle physics produce enormous quantities of data that must be analyzed and interpreted by teams of physicists. This analysis is often exploratory, where scientists are unable to enumerate the possible types of signal prior to performing the experiment. Thus, tools for summarizing, clustering, visualizing and classifying high-dimensional data are essential. In this work, we show that meaningful physical content can be revealed by transforming the raw data into a learned high-level representation using deep neural networks, with measurements taken at the Daya Bay Neutrino Experiment as a case study. We further show how convolutional deep neural networks can provide an effective classification filter with greater than 97% accuracy across different classes of physics events, significantly better than other machine learning approaches.

Recent progress in deep learning algorithms for artificial intelligence has raised widespread ethical concerns [1][2]. It has been argued that human-level AI isn't automatically good for humanity. It might be presumptuous and overconfident to be sure that humans would be able to control superhuman-clever AIs, that those AIs would really care about humans, for example to allow us full access to mineral resources and agriculture fields of the planet. While there are numerous advantages of having clever AIs in the short-term, the long-term danger of having too clever AIs might outweigh, leading to net negative effect of AI progress on society. The most common argument against consideration of such long-term risks is their vagueness due to supposed very long time distance from us [3].

The call for proposals is open for the O'Reilly Artificial Intelligence Conference NY 2017 now through January 18, 2017. Subscribe to the O'Reilly Data Show Podcast to explore the opportunities and techniques driving big data, data science, and AI. Find us on Stitcher, TuneIn, iTunes, SoundCloud, RSS. In this episode I spoke with Vikash Mansinghka, research scientist at MIT, where he leads the Probabilistic Computing Project, and co-founder of Empirical Systems. I've long wanted to introduce listeners to recent developments in probabilistic programming, and I found the perfect guide in Mansinghka.

Uber is coming closer to its self-driving cars goal. The ride-sharing company announced Monday it launched Uber AI Labs, a new department based in San Francisco which will focus on research in artificial intelligence and machine learning. Uber also revealed it has acquired the AI research startup Geometric Intelligence. The 15 members of the startup will form the primary core of the AI Labs team. With the company's new moves, Uber is betting that artificial intelligence could improve its services.

But when is the last time you saw someone building a house with a hammer, a hand saw and some 2x4s? When you build a house, you need the right tools and materials to build a house. The basic ingredients for learning are neurons and myelin. Each time you fire a set of neurons while learning, they get wrapped in another thin layer of myelin, which is like insulation on an electric cord. The more the neurons get wrapped up, the faster the neurons can send signals.

I am a data scientist, big data engineer, and full stack software engineer. For my masters thesis I worked on brain-computer interfaces using machine learning. These assist non-verbal and non-mobile persons communicate with their family and caregivers. I have worked in online advertising and digital media as both a data scientist and big data engineer, and built various high-throughput web services around said data. I've created new big data pipelines using Hadoop/Pig/MapReduce. I've created machine learning models to predict click-through rate, news feed recommender systems using linear regression, Bayesian Bandits, and collaborative filtering and validated the results using A/B testing.