We'll explore what neural networks are, how they work, and how they're used today in today's rapidly developing machine-learning world. Before we look at different types of neural networks, we need to start with the basic building blocks.

A research collaboration has achieved exaflop performance on the Summit supercomputer with a deep learning application used to model subsurface flow in the study of nuclear waste remediation. Their achievement, which will be presented during the "Deep Learning on Supercomputers" workshop at SC19, demonstrates the promise of physics-informed generative adversarial networks (GANs) for analyzing complex, large-scale science problems. The concept of physics-informed GANs is to encode prior information from physics into the neural network. This allows you to go well beyond the training domain, which is very important in applications where the conditions can change. GANs have been applied to model human face appearance with remarkable accuracy.

Such is computing's future--to each of our voices. Voice assistants hang on every word we say, when prompted. Their genesis has created an entire family named Siri, Alexa, Cortana, Pepper, Watson, and, most recently, Duplex. One wonders when smart speakers will replace what you're looking at now--this screen. The promise of voice interfaces was demonstrated again in May when Google rolled out Duplex, an AI-driven voice assistant so lifelike and sophisticated that some found it astounding but others deemed it unsettling and wondered if a new era of robocall abuse just dawned.

We are living in an era that is showing massive growth in data and computing power. We have seen a lot of progress in machine learning and deep learning, but there is an ever-growing need for more intelligent, more explainable AI. Most people's perception of artificial intelligence boils down to either science fiction, or what we call conventional AI. The foundations of conventional AI are numerical techniques like data analytics, including statistical analysis, modeling, and machine learning. This has been the primary approach to AI over the past few decades with significant success by numerous companies in many industries.

One of the biggest challenges in machine learning workflows is identifying which inputs in your data will provide the best signals for training predictive models. For image data and other unstructured formats, deep learning models are showing large improvements over prior approaches, but for data already in structured formats, the benefits are less obvious. At Zynga, I've been exploring feature generation methods for shallow learning problems, where our data is already in a structured format, and the challenge is to translate thousands of records per user into single records that summarize user activity. Once you have the ability to translate raw tracking events into user summaries, you can apply a variety of supervised and unsupervised learning methods to your application. I've been leveraging the Featuretools library to significantly reduce my time spent building predictive models, and it's unlocked a new class of problems that data scientists can address.

Udacity, the Silicon Valley based lifelong learning platform, announced its newest initiative to expand students' artificial intelligence skills: the Intel Edge AI Scholarship Program. This new scholarship program, announced at the Intel AI Summit and the Future of Education and Workforce Summit in San Francisco, will empower professional developers interested in advanced learning, specifically deep learning and computer vision, to accelerate the development and deployment of high-performance computer vision and deep learning solutions. Computer vision and AI at the edge are becoming instrumental in powering everything from factory assembly lines and retail inventory management to hospital urgent care medical imaging equipment like X-ray and CAT scans. This program will teach fluency in some of the most cutting-edge technologies. Upon successful completion of the first phase of the program, students will also have the opportunity to earn their way to a full scholarship to the Intel Edge AI for IoT Developers Nanodegree program, a brand-new Udacity Nanodegree program built in partnership with Intel.

Long read sequencing technologies such as Oxford Nanopore can greatly decrease the complexity of de novo genome assembly and large structural variation identification. Currently Nanopore reads have high error rates, and the errors often cluster into low-quality segments within the reads. The limited sensitivity of existing read-based error correction methods can cause large-scale mis-assemblies in the assembled genomes, motivating further innovation in this area. Here we developed a Convolutional Neural Network (CNN) based method, called MiniScrub, for identification and subsequent "scrubbing" (removal) of low-quality Nanopore read segments to minimize their interference in downstream assembly process. MiniScrub first generates read-to-read overlaps via MiniMap2, then encodes the overlaps into images, and finally builds CNN models to predict low-quality segments.

Generative adversarial networks (GANs) are initially proposed to generate images by learning from a large number of samples. Recently, GANs have been used to emulate complex physical systems such as turbulent flows. However, a critical question must be answered before GANs can be considered trusted emulators for physical systems: do GANs-generated samples conform to the various physical constraints? These include both deterministic constraints (e.g., conservation laws) and statistical constraints (e.g., energy spectrum in turbulent flows). The latter have been studied in a companion paper (Wu et al. 2019. In the present work, we enforce deterministic yet approximate constraints on GANs by incorporating them into the loss function of the generator. We evaluate the performance of physics-constrained GANs on two representative tasks with geometrical constraints (generating points on circles) and differential constraints (generating divergence-free flow velocity fields), respectively. In both cases, the constrained GANs produced samples that precisely conform to the underlying constraints, even though the constraints are only enforced approximately. More importantly, the imposed constraints significantly accelerate the convergence and improve the robustness in the training. These improvements are noteworthy, as the convergence and robustness are two well-known obstacles in the training of GANs. Keywords: Generative adversarial networks, physics constraints, physics-informed machine learning 1. Introduction Machine learning and particularly deep learning has achieved significant success in a wide range of commercial domain applications such as image recognition, audio recognition, and natural language processing [1-5]. Corresponding author Email address: hengxiao@vt.edu For example, machine learning methods such as random forests and neural networks have been used to provide closure models for turbulent flows [6-9] and multiphase flows [10, 11] and to compute rock permeability directly from CT scan images [12]. They have also been used to discover ordinary and partial differential equations (ODEs and PDEs) from data [13-16]. Finally, neural networks have been used to solve exactly specified PDEs [17-20] and partially known PDEs by incorporating available data [21-24]. The scientific applications reviewed above mostly involve supervised learning, which consists of three steps: (a) postulate a model that maps inputs (features) to outputs (labels), controlled by a set of adjustable model parameters; (b) learn the parameters from training data (labeled examples of input-output pairs); and (c) use the fitted model to predict the responses for new inputs that were not included in the training data.

Most distortion correction methods focus on simple forms of distortion, such as radial or linear distortions. These works undistort images either based on measurements in the presence of a calibration grid, or use multiple views to find point correspondences and predict distortion parameters. When possible distortions are more complex, e.g. in the case of a camera being placed behind a refractive surface such as glass, the standard method is to use a calibration grid. Considering a high variety of distortions, it is nonviable to conduct these measurements. In this work, we present a single view distortion correction method which is capable of undistorting images containing arbitrarily complex distortions by exploiting recent advancements in differentiable image sampling and in the usage of semantic information to augment various tasks. The results of this work show that our model is able to estimate and correct highly complex distortions, and that incorporating semantic information mitigates the process of image undistortion.

Recent work has shown that deep generative models assign higher likelihood to out-of-distribution inputs than to training data. We show that a factor underlying this phenomenon is a mismatch between the nature of the prior distribution and that of the data distribution, a problem found in widely used deep generative models such as VAEs and Glow. While a typical choice for a prior distribution is a standard Gaussian distribution, properties of distributions of real data sets may not be consistent with a unimodal prior distribution. This paper focuses on the relationship between the choice of a prior distribution and the likelihoods assigned to out-of-distribution inputs. We propose the use of a mixture distribution as a prior to make likelihoods assigned by deep generative models sensitive to out-of-distribution inputs. Furthermore, we explain the theoretical advantages of adopting a mixture distribution as the prior, and we present experimental results to support our claims. Finally, we demonstrate that a mixture prior lowers the out-of-distribution likelihood with respect to two pairs of real image data sets: Fashion-MNIST vs. MNIST and CIFAR10 vs. SVHN.