Search

Showing total 4 results
Start Over Topic artificial neural networks Publication Year Range Last 10 years
4 results

1. A Single-Shot Generalized Device Placement for Large Dataflow Graphs.

Author

Zhou, Yanqi, Roy, Sudip, Abdolrashidi, Amirali, Wong, Daniel Lin-Kit, Ma, Peter, Xu, Qiumin, Mirhoseini, Azalia, and Laudon, James

Subjects
*RECURRENT neural networks, *ARTIFICIAL neural networks, *DEEP learning
Abstract

With increasingly complex neural network architectures and heterogeneous device characteristics, finding a reasonable graph partitioning and device placement strategy is challenging. There have been prior attempts at learned approaches for solving device placement, these approaches are computationally expensive, unable to handle large graphs consisting over 50000 nodes, and do not generalize well to unseen graphs. To address all these limitations, we propose an efficient single-shot, generalized deep RL method (SGDP) based on a scalable sequential attention mechanism over a graph neural network that is transferable to new graphs. On a diverse set of representative deep learning models, our method on average achieves 20% improvement over human placement and 18% improvement over the prior art with 15× faster convergence. We are the first to demonstrate super human performance on 8-layer recurrent neural network language model and 8-layer GNMT consisting of over 50000 nodes, on 8-GPUs. We provide rationales and sensitivity study on model architecture selections. [ABSTRACT FROM AUTHOR]

Published
2020
Full Text
View/download PDF

2. Application of Neural Networks for Estimation of Paper Properties Based on Refined Pulp Properties.

Author

Ciesielski, Krzysztof and Olejnik, Konrad

Subjects
PAPER industry, ARTIFICIAL neural networks, PRODUCT quality, PAPERMAKING, DECISION making, TENSILE strength
Abstract

Copyright of Fibres & Textiles in Eastern Europe is the property of Sciendo and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Published
2014

3. Machine learning for direct oxygen saturation and hemoglobin concentration assessment using diffuse reflectance spectroscopy

Author

Ingemar Fredriksson, Marcus Larsson, and Tomas Strömberg

Subjects
Paper, Materials science, Diffuse reflectance infrared fourier transform, Medical Laboratory and Measurements Technologies, Biomedical Engineering, Analytical chemistry, chemistry.chemical_element, Special Series on Artificial Intelligence and Machine Learning in Biomedical Optics, microcirculation, 01 natural sciences, Oxygen, Noise (electronics), law.invention, Monte Carlo simulations, diffuse reflectance spectroscopy, 010309 optics, Biomaterials, Machine Learning, Hemoglobins, law, 0103 physical sciences, Calibration, Medicinsk laboratorie- och mätteknik, Oxygen saturation (medicine), Phantoms, Imaging, Spectrum Analysis, hemoglobin oxygen saturation, multilayer tissue model, Atomic and Molecular Physics, and Optics, Electronic, Optical and Magnetic Materials, Wavelength, artificial neural networks, Pressure measurement, chemistry, Attenuation coefficient
Abstract

Significance: Diffuse reflectance spectroscopy (DRS) is frequently used to assess oxygen saturation and hemoglobin concentration in living tissue. Methods solving the inverse problem may include time-consuming nonlinear optimization or artificial neural networks (ANN) determining the absorption coefficient one wavelength at a time. Aim: To present an ANN-based method that directly outputs the oxygen saturation and the hemoglobin concentration using the shape of the measured spectra as input. Approach: A probe-based DRS setup with dual source-detector separations in the visible wavelength range was used. ANNs were trained on spectra generated from a three-layer tissue model with oxygen saturation and hemoglobin concentration as target. Results: Modeled evaluation data with realistic measurement noise showed an absolute root-mean-square (RMS) deviation of 5.1% units for oxygen saturation estimation. The relative RMS deviation for hemoglobin concentration was 13%. This accuracy is at least twice as good as our previous nonlinear optimization method. On blood-intralipid phantoms, the RMS deviation from the oxygen saturation derived from partial oxygen pressure measurements was 5.3% and 1.6% in two separate measurement series. Results during brachial occlusion showed expected patterns. Conclusions: The presented method, directly assessing oxygen saturation and hemoglobin concentration, is fast, accurate, and robust to noise. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Funding Agencies|Swedish Research CouncilSwedish Research Council [2014-6141]; Swedens innovation agency VINNOVA via the program MedTech4Health [2016-02211]; Swedens innovation agency VINNOVA via the program Swelife and MedTech4Health [2017-01435]

Published
2020

4. Machine learning in multiexposure laser speckle contrast imaging can replace conventional laser Doppler flowmetry

Author

Ingemar, Fredriksson, Martin, Hultman, Tomas, Strömberg, and Marcus, Larsson

Subjects
Adult, Male, Paper, Stochastic Processes, Erythrocytes, Models, Statistical, laser speckle contrast analysis, Lasers, Microcirculation, Reproducibility of Results, Imaging, Machine Learning, Perfusion, Regional Blood Flow, Calibration, Image Processing, Computer-Assisted, Laser-Doppler Flowmetry, Humans, blood flow, Computer Simulation, Neural Networks, Computer, Monte Carlo Method, artificial neural networks, Blood Flow Velocity
Abstract

Laser speckle contrast imaging (LSCI) enables video rate imaging of blood flow. However, its relation to tissue blood perfusion is nonlinear and depends strongly on exposure time. By contrast, the perfusion estimate from the slower laser Doppler flowmetry (LDF) technique has a relationship to blood perfusion that is better understood. Multiexposure LSCI (MELSCI) enables a perfusion estimate closer to the actual perfusion than that using a single exposure time. We present and evaluate a method that utilizes contrasts from seven exposure times between 1 and 64 ms to calculate a perfusion estimate that resembles the perfusion estimate from LDF. The method is based on artificial neural networks (ANN) for fast and accurate processing of MELSCI contrasts to perfusion. The networks are trained using modeling of Doppler histograms and speckle contrasts from tissue models. The importance of accounting for noise is demonstrated. Results show that by using ANN, MELSCI data can be processed to LDF perfusion with high accuracy, with a correlation coefficient R = 1.000 for noise-free data, R = 0.993 when a moderate degree of noise is present, and R = 0.995 for in vivo data from an occlusion-release experiment.

Published
2018