Search NASA⌕ Search

Engineering topics

Young, Steven

Publications and source records attributed to Young, Steven.

At least 37 records · Page 2

Visualization System for Evolutionary Neural Networks for Deep Learning

Deep learning is actively used in a wide range of fields for scientific discovery. To effectively apply deep learning to a particular problem, it is important to select an appropriate network architecture and other hyper-parameters (at each layer). Evolving architectures and hyper-parameters using a genetic algorithm is one current approach to search the huge space of all possible configurations to find those more optimal for the problem. However, examining an evolutionary process and tuning the genetic algorithm are challenging, pushing most users to treat the process as a black box. To address this challenge, we propose a visualization system for evolutionary neural networks for deep learning. The key feature of our visualization system is to provide a visual analytics environment for evaluating a genetic algorithm in order to improve the underlying operations to reduce time to find good solutions. Our system is able to not only visualize how a genetic algorithm traverses its search space but also allows users to examine evolving networks in-depth to get insights to improve performance through interactive visualization components.

Chae, Junghoon↗

Inferring Convolutional Neural Networks' Accuracies from Their Architectural Characterizations

The challenge of choosing an appropriate convolutional neural network (CNN) architecture for specific applications and different data sets is still poorly understood in the literature. This is problematic, since CNNs have shown strong promise for analyzing scientific data from many domains including particle imaging detectors. In this paper, we proposed a systematic language that is useful for comparison between different CNN's architectures before training time. This helps us predict whether a network can perform better than a certain threshold accuracy before training up to 70% accuracy using simple machine learning models. Additionally, we found a coefficient of determination of 0.966 for an Ordinary Least Squares model in a regression task to predict accuracy of a large population of networks.

Hoang, Duc↗

Fine-Grained Exploitation of Mixed Precision for Faster CNN Training

As deep convolutional neural networks (CNNs) have become increasingly popular and successful at an ever-widening number of machine learning tasks specialized hardware has become increasingly available for training and deploying them. NVIDIA's recent Volta architecture includes tensor cores which perform a fused operation reduced and mixed precision (16-bit multiply, 32-bit accumulate). Recent research indicates that, typically, very little is lost (in terms of training accuracy) when half precision is used in place of single precision, and performance gains can be made by doing arithmetic in reduced precision. In this work we demonstrate that making layer-by-layer choices as to the arithmetic/data precision can lead to further performance improvement. In our study of 25,200 CNNs we demonstrate an average speedup (over purely half precision) of 1.27x and speedups as high as 3.64x by appropriately combining single and half precision arithmetic and data types on a layer-by-layer basis.

Johnston, Travis↗

Reinforcement Learning-based Traffic Control to Optimize Energy Usage and Throughput (CRADA report)

The US roadways are critical to meeting the mobility and economic needs of the nation. The United States uses 28% of its energy in moving goods and people, with approximately 60% of that utilized by cars, light trucks, and motorcycles. Thus, improved transportation efficiency is vital to America’s economic progress. The increasing congestion and energy resource requirements of transportation systems for metropolitan areas require research in methods to improve and optimize control methods. Coordinating and optimizing traffic in urban areas may introduce hundreds of thousands of vehicles and traffic management systems, which can require high performance computing (HPC) resources to model and manage. In this work, we seek to use machine learning, computer vision, and HPC to improve the energy efficiency aspects of traffic control by leveraging GRIDSMART traffic cameras as sensors for adaptive traffic control, with a sensitivity to the fuel consumption characteristics of the traffic in the camera’s visual field. Traffic control use cases using reinforcement learning have been published and achieved good results. Surveys from DOE national laboratories estimate that the fuel cost of idling is 6 billion gallons wasted annually [Argonne National Laboratory, 2019]. GRIDSMART cameras—an existing, fielded commercial product—sense the presence of vehicles at intersections and replace more conventional sensors (such as inductive loops) to issue calls to traffic control. These cameras, which have horizon-to-horizon view, offer the potential for an improved view of the traffic environment which can be used to generate better control algorithms.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Self-Taught Waveform Synthesis and Analysis in the Amplify-and-Forward Relay Channel

Wireless communications plays a pivotal role in multiple complex domains such as tactical networks or space communications. Traditional physical (PHY) layer protocols for digital communications contain chains of signal processing blocks that have been mathematically optimized to transmit information bits efficiently over noisy channels. Unfortunately, the ongoing advancement of hardware and software design, and algorithm development, makes it difficult for some domains to keep up with the constant change in modern communication systems. It has been shown previously that combining deep learning with digital modulation (deepmod) allows a system to learn communications on its own rather than requiring human-invented protocols. This is particularly attractive to space communications where updating PHY layer technologies may be prohibitively complex or expensive. A link using deepmod is able to learn both waveform synthesis (transmit) and analysis (receive) that is self-taught. When deepmod is first initiated it has no knowledge of the channel medium but quickly learns to communicate by synthesizing waveforms that can be successfully decoded at the other end of the link. This is accomplished by a custom deep neural network especially suited for this particular task of learning. In this current work, we show that deepmod learns in both traditional point-to-point channels as well as the more abstract multi-hop amplify-and-forward relay channel. In the experimental results, even though no direct link between transmitter and receiver exists, deepmod-enabled nodes still create latent information bearing waveforms that can be used for communications.

Anderson, Adam L.↗

167-PFlops deep learning for electron microscopy: from learning physics to atomic manipulation

An artificial intelligence system called MENNDL, which used 25,200 NVIDIA Volta GPUs on Oak Ridge National Laboratory's Summit machine, automatically designed an optimal deep learning network in order to extract structural information from raw atomic-resolution microscopy data. In a few hours, MENNDL creates and evaluates millions of networks using a scalable, parallel, asynchronous genetic algorithm augmented with a support vector machine to automatically find a superior deep learning network topology and hyper-parameter set than a human expert can find in months. For the application of electron microscopy, the system furthers the goal of improving our understanding of the electron-beam-matter interactions and real-time image-based feedback, which enables a huge step beyond human capacity towards nanofabricating materials automatically. MENNDL has been scaled to the 4,200 available nodes of Summit achieving a measured 152.5 PFlops, with an estimated sustained performance of 167 PFlops when the entire machine is available.

Patton, Robert↗

Deepmod: An Over-the-Air Trainable Machine Modem for Resilient PHY Layer Communications

Traditional physical layer protocols (e.g. WiFi, WiMax, etc.) are well established and are often optimal in a wide variety of channel conditions including heterogenous links and in tactical communications. Unfortunately, this same optimality encourages ubiquity in wireless communication technology and enhances the potential for catastrophic cyber or physical attacks due to prolific knowledge of underlying physical layers. Any truly resilient communications protocol must be capable of immediate redeployment to meet quality of service (QoS) demands in a wide variety of possible channel media. This work proposes an approach to communications that is contrary to much traditional approaches in that processing blocks are generated real-time and only relevant to the particular channel medium being used. Rather than creating man-made ubiquitous blocks of signal processing, we examine using processing that is immediately expendable once it has been used. This is achieved through software-defined radios, and deep modulation, where system blocks are replaced with machine learning graphs that can be trained, used, and then discarded as needed. Simulation and experimental hardware show how deep modulation can converge to viable communications links, using the same machine intelligence, in vastly different channels.

Anderson, Adam L↗

Unsupervised Identification of Study Descriptors in Toxicology Research: An Experimental Study

Identifying and extracting data elements such as study descriptors in publication full texts is a critical yet manual and labor-intensive step required in a number of tasks. In this paper we address the question of identifying data elements in an unsupervised manner. Specifically, provided a set of criteria describing specific study parameters, such as species, route of administration, and dosing regimen, we develop an unsupervised approach to identify text segments relevant to the criteria. A binary classifier trained to identify publications that met the criteria performs better when trained on the candidate sentences than when trained on sentences randomly picked from the text, supporting the intuition that our method is able to accurately identify study descriptors.

Herrmannova, Dasha↗

A Study of Complex Deep Learning Networks on High-Performance, Neuromorphic, and Quantum Computers

Current deep learning approaches have been very successful using convolutional neural networks trained on large graphical-processing-unit-based computers. Three limitations of this approach are that (1) they are based on a simple layered network topology, i.e., highly connected layers, without intra-layer connections; (2) the networks are manually configured to achieve optimal results, and (3) the implementation of the network model is expensive in both cost and power. In this paper, we evaluate deep learning models using three different computing architectures to address these problems: quantum computing to train complex topologies, high performance computing to automatically determine network topology, and neuromorphic computing for a low-power hardware implementation. We use the MNIST dataset for our experiment, due to input size limitations of current quantum computers. Our results show the feasibility of using the three architectures in tandem to address the above deep learning limitations. Finally, we show that a quantum computer can find high quality values of intra-layer connection weights in a tractable time as the complexity of the network increases, a high performance computer can find optimal layer-based topologies, and a neuromorphic computer can represent the complex topology and weights derived from the other architectures in low power memristive hardware.

97 MATHEMATICS AND COMPUTING↗

Adiabatic Quantum Computation Applied to Deep Learning Networks

Training deep learning networks is a difficult task due to computational complexity, and this is traditionally handled by simplifying network topology to enable parallel computation on graphical processing units (GPUs). However, the emergence of quantum devices allows reconsideration of complex topologies. We illustrate a particular network topology that can be trained to classify MNIST data (an image dataset of handwritten digits) and neutrino detection data using a restricted form of adiabatic quantum computation known as quantum annealing performed by a D-Wave processor. We provide a brief description of the hardware and how it solves Ising models, how we translate our data into the corresponding Ising models, and how we use available expanded topology options to explore potential performance improvements. Although we focus on the application of quantum annealing in this article, the work discussed here is just one of three approaches we explored as part of a larger project that considers alternative means for training deep learning networks. The other approaches involve using a high performance computing (HPC) environment to automatically find network topologies with good performance and using neuromorphic computing to find a low-power solution for training deep learning networks. Our results show that our quantum approach can find good network parameters in a reasonable time despite increased network topology complexity; that HPC can find good parameters for traditional, simplified network topologies; and that neuromorphic computers can use low power memristive hardware to represent complex topologies and parameters derived from other architecture choices.

97 MATHEMATICS AND COMPUTING↗

Neural Networks and Graph Algorithms with Next-Generation Processors

The use of graphical processors for distributed computation revolutionized the field of high performance scientific computing. As the Moore's Law era of computing draws to a close, the development of non-Von Neumann systems: neuromorphic processing units, and quantum annealers; again are redefining new territory for computational methods. While these technologies are still in their nascent stages, we discuss their potential to advance computing in two domains: machine learning, and solving constraint satisfaction problems. Each of these processors utilize fundamentally different theoretical models of computation. This raises questions about how to best use them in the design and implementation of applications. While many processors are being developed with a specific domain target, the ubiquity of spin-glass models and neural networks provides an avenue for multi-functional applications. This provides hints at the future infrastructure needed to integrate many next-generation processing units into conventional high-performance computing systems.

Hamilton, Kathleen↗

DeepPDF: A Deep Learning Approach to Extracting Text from PDFs

Scientific publications contain a plethora of important information, not only for researchers but also for their managers and institutions. Many researchers try to collect and extract this information in large enough quantities that it requires machine automation, but because publications were historically intended for print and not machine consumption, the digital document formats used today (primarily PDF) have created many hurdles for text extraction. Primarily, tools have relied on trying to convert PDF's to plain text documents for machine processing by reverse engineering the PDF standard. A complex process because once a PDF is created it is more closely related to an image file than a document markup language. In this paper we explore the feasibility of treating these PDF documents as images as opposed to a proprietary markup language. We believe that by using deep learning and image analysis we can create more accurate PDF to text extraction tools than those that currently exist. \\ \newline \Keywords{deep learning, text extraction, information extraction, PDF extraction, scholarly publications.

Stahl, Christopher↗

Neuromorphic computing for temporal scientific data classification

In this work, we apply a spiking neural network model and an associated memristive neuromorphic implementation to an application in classifying temporal scientific data. We demonstrate that the spiking neural network model achieves comparable results to a previously reported convolutional neural network model, with significantly fewer neurons and synapses required.

Schuman, Catherine↗

Evolving Deep Networks Using HPC

While a large number of deep learning networks have been studiedand published that produce outstanding results on natural imagedatasets, these datasets only make up a fraction of those to whichdeep learning can be applied. These datasets include text data, audiodata, and arrays of sensors that have very different characteristicsthan natural images. As these “best” networks for natural imageshave been largely discovered through experimentation and cannotbe proven optimal on some theoretical basis, there is no reasonto believe that they are the optimal network for these drasticallydifferent datasets. Hyperparameter search is thus often a very im-portant process when applying deep learning to a new problem. Inthis work we present an evolutionary approach to searching thepossible space of network hyperparameters and construction thatcan scale to 18, 000 nodes. This approach is applied to datasets ofvarying types and characteristics where we demonstrate the abilityto rapidly find best hyperparameters in order to enable practitionersto quickly iterate between idea and result.

Young, Steven↗

Vertex reconstruction of neutrino interactions using deep learning

Deep learning offers new tools to improve our understanding of many important scientific problems. Neutrinos are the most abundant particles in existence and are hypothesized to explain the matter-antimatter asymmetry that dominates our universe. Definitive tests of this conjecture require a detailed understanding of neutrino interactions with a variety of nuclei. Many measurements of interest depend on vertex reconstruction -- finding the origin of a neutrino interaction using data from the detector, which can be represented as images. Traditionally, this has been accomplished by utilizing methods that identify the tracks coming from the interaction. However, these methods are not ideal for interactions where an abundance of tracks and cascades occlude the vertex region. Manual algorithm engineering to handle these challenges is complicated and error prone. Deep learning extracts rich, semantic features directly from raw data, making it a promising solution to this problem. In this work, deep learning models are presented that classify the vertex location in regions meaningful to the domain scientists improving their ability to explore more complex interactions.

Terwilliger, Adam↗

Emerging Technologies for Airplane State Awareness and Prediction

Loss of control in flight (LOC-I) is consistently the leading cause of fatal aircraft accidents. A study of LOC accidents and incidents, commissioned by the Commercial Aviation Safety Team (CAST) identified a growing trend in loss of Airplane State Awareness (ASA) by the flight crew. This has led to recommended safety enhancements that include flight deck technologies with the potential of enhancing flight crew awareness of airplane energy state. The goal of this research is to develop and evaluate technologies that predict and assess the future aircraft energy state and auto-flight configuration, and provide appropriate alerting to anticipated problematic auto-flight inputs, with the aim of enhancing pilots situational awareness.

Aircraft Safety↗