scholarly journals Conservation laws in a neural network architecture: Enforcing the atom balance of a Julia-based photochemical model (v0.2.0)

2021 ◽  
Author(s):  
Patrick Obin Sturm ◽  
Anthony S. Wexler
Keyword(s):  
Neural Network ◽  
Conservation Laws ◽  
Network Architecture ◽  
Neural Network Architecture ◽  
Machine Precision ◽  
Emissions Scenarios ◽  
Hidden Layer ◽  
Computational Resources ◽  
Atmospheric Phenomena ◽  
Insight Into

Abstract. Models of atmospheric phenomena provide insight into climate, air quality, and meteorology, and provide a mechanism for understanding the effect of future emissions scenarios. To accurately represent atmospheric phenomena, these models consume vast quantities of computational resources. Machine learning (ML) techniques such as neural networks have the potential to emulate compute-intensive components of these models to reduce their computational burden. However, such ML surrogate models may lead to nonphysical predictions that are difficult to uncover. Here we present a neural network architecture that enforces conservation laws. Instead of simply predicting properties of interest, a physically interpretable hidden layer within the network predicts fluxes between properties which are subsequently related to the properties of interest. As an example, we design a physics-constrained neural network surrogate model of photochemistry using this approach and find that it conserves atoms as they flow between molecules to machine precision, while outperforming a naïve neural network in terms of accuracy and non-negativity of concentrations.

scholarly journals Image reconstruction through a multimode fiber with a simple neural network architecture

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Changyan Zhu ◽  
Eng Aik Chan ◽  
You Wang ◽  
Weina Peng ◽  
Ruixiang Guo ◽  
...  
Keyword(s):  
Neural Network ◽  
Image Reconstruction ◽  
Network Architecture ◽  
Neural Network Architecture ◽  
Training Time ◽  
Modal Dispersion ◽  
Multimode Fibers ◽  
Speckle Patterns ◽  
Hidden Layer ◽  
Simple Neural Network

AbstractMultimode fibers (MMFs) have the potential to carry complex images for endoscopy and related applications, but decoding the complex speckle patterns produced by mode-mixing and modal dispersion in MMFs is a serious challenge. Several groups have recently shown that convolutional neural networks (CNNs) can be trained to perform high-fidelity MMF image reconstruction. We find that a considerably simpler neural network architecture, the single hidden layer dense neural network, performs at least as well as previously-used CNNs in terms of image reconstruction fidelity, and is superior in terms of training time and computing resources required. The trained networks can accurately reconstruct MMF images collected over a week after the cessation of the training set, with the dense network performing as well as the CNN over the entire period.

FEEDFORWARD NEURAL NETWORK WITH ADAPTIVE REFERENCE PATTERN LAYER

1999 ◽  
Vol 09 (01) ◽  
pp. 1-9
Author(s):  
MIKKO LEHTOKANGAS
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Neural Network Architecture ◽  
Reference Pattern ◽  
Hybrid Neural Network ◽  
Rbf Networks ◽  
Pattern Layer ◽  
Input Variables ◽  
Hidden Layer ◽  
Efficient Learning

A hybrid neural network architecture is investigated for modeling purposes. The proposed hybrid is based on the multilayer perceptron (MLP) network. In addition to the usual hidden layers, the first hidden layer is selected to be an adaptive reference pattern layer. Each unit in this new layer incorporates a reference pattern that is located somewhere in the space spanned by the input variables. The outputs of these units are the component wise-squared differences between the elements of a reference pattern and the inputs. The reference pattern layer has some resemblance to the hidden layer of the radial basis function (RBF) networks. Therefore the proposed design can be regarded as a sort of hybrid of MLP and RBF networks. The presented benchmark experiments show that the proposed hybrid can provide significant advantages over standard MLPs and RBFs in terms of fast and efficient learning, and compact network structure.

scholarly journals Identifikasi Penyakit Diabetes Millitus Menggunakan Jaringan Syaraf Tiruan Dengan Metode Perambatan-Balik (Backpropagation)

2018 ◽  
Author(s):  
Sutedi Sutedi
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Network Architecture ◽  
Neural Network Architecture ◽  
Accuracy Rate ◽  
Layer 2 ◽  
Input Layer ◽  
Hidden Layer ◽  
Diabetes Melitus

Diabetes Melitus (DM) is dangerous disease that affect many of the variouslayer of work society. This disease is not easy to accurately recognized by thegeneral society. So we need to develop a system that can identify accurately. Systemis built using neural networks with backpropagation methods and the functionactivation sigmoid. Neural network architecture using 8 input layer, 2 output layerand 5 hidden layer. The results show that this methods succesfully clasifies datadiabetics and non diabetics with near 100% accuracy rate.

LESIONING MCCLOSKEY AND LINDEMANN’S (1992) MATHNET: THE EFFECT OF DAMAGE LOCATION AND AMOUNT

1994 ◽  
Vol 02 (03) ◽  
pp. 335-356 ◽  
Author(s):  
G. LORIES ◽  
A. AUBRUN ◽  
X. SERON
Keyword(s):  
Neural Network ◽  
Brain Damage ◽  
Network Architecture ◽  
Error Pattern ◽  
Neural Network Architecture ◽  
Damage Location ◽  
Error Distributions ◽  
Fact Retrieval ◽  
Insight Into ◽  
Simple Network

McCloskey and Lindemann [32] provide a simulation of brain damage on a neural network architecture and offer evidence that different lesions to a same network can lead to different error distributions. We briefly review the various kinds of networks that have been proposed to simulate various arithmetical fact retrieval phenomena and we present a simple network designed to make some computational constraints apparent. Additionally, we replicate McCloskey and Lindemann’s [32] simulation by training 5 different artificial “subjects” and inflicting various types of damage upon each. Examination of the behaviour of our version of the network after different amounts of damage to its various connection blocks confirms that the error pattern may vary. These variations in the error patterns can be analyzed. The data may help to clarify the functioning of the network and give insight into the reasons why it produces several effects observed in human behaviour.

scholarly journals PREDICTION OF THE SIZE OF NANOPARTICLES AND MICROSPORE SURFACE AREA USING ARTIFICIAL NEURAL NETWORK

2018 ◽  
Vol 1 (1) ◽  
pp. 65
Author(s):  
Dženana Sarajlić ◽  
Layla Abdel-Ilah ◽  
Adnan Fojnica ◽  
Ahmed Osmanović
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Transfer Function ◽  
Surface Area ◽  
Network Architecture ◽  
Neural Network Architecture ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Hidden Layer ◽  
Experimental Values

This paper presents development of Artificial Neural Network (ANN) for prediction of the size of nanoparticles (NP) and microspore surface area (MSA). Developed neural network architecture has the following three inputs: the concentration of the biodegradable polymer in the organic phase, surfactant concentration in the aqueous phase and the homogenizing pressure. Two-layer feedforward network with a sigmoid transfer function in the hidden layer and a linear transfer function in the output layer is trained, using Levenberg-Marquardt training algorithm. For training of this network, as well as for subsequent validation, 36 samples were used. From 36 samples which were used for subsequent validation in this ANN, 80,5% of them had highest accuracy while 19,5% of output data had insignificant differences comparing to experimental values.

scholarly journals PERAMALAN JUMLAH PERMINTAAN UDANG BEKU PND MENGGUNAKAN METODE JARINGAN SYARAF TIRUAN (JST) BACKPROPAGATION

2017 ◽  
Vol 11 (1) ◽  
pp. 17 ◽  
Author(s):  
Iid Mufidah ◽  
Sony Suwasono ◽  
Yuli Wibowo ◽  
Deddy Wirawan Soedibyo
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Network Architecture ◽  
Series Data ◽  
Architecture Design ◽  
Neural Network Architecture ◽  
Output Layer ◽  
Artificial Neural Network Architecture ◽  
Artificial Neural ◽  
Hidden Layer

Forecasting is the art or science to estimate how many needs will come in order to meet the demand for goods or services, often based on historical time series data. The growing number of emerging companies in Indonesia today has created a very tight business competition in both services and products. Consumers choose the best service and high quality and low price. Consumer demand is always uncertain or varied in each subsequent period. The aim of this research was to determind the best backpropagation neural network architecture design and to predict the demand of frozen product of PND 26/30. This research used the method of Neural Network (ANN) and Processing ANN using MATLAB software. Implementation of ANN method in PT.XYZ using Backpropagation algorithm. Artificial neural network architecture used was 12 input layer, 1 output layer, and 12 hidden layer and activation function used tansig and purelin. Tansig for hidden layer and purelin for output layer. The best artificial neural network architecture design for product demand for PND 31/40 was a multi layer feedforward value of Mean Square Error (MSE) network training value of 0.01 with MAPE 3.35. The result of JST forecasting period 2017 were 960 MC, 637 MC, 572 MC, 993 MC, 1386 MC, 480 MC, 135 MC, 1209 MC, 1476 MC, 1029 MC, 290 MC, and 952 MC. Keywords: artificial neural network, PND 26/30, backpropagation, MSE, MAPE

scholarly journals TAPAS: Train-Less Accuracy Predictor for Architecture Search

2019 ◽  
Vol 33 ◽  
pp. 3927-3934 ◽  
Author(s):  
R. Istrate ◽  
F. Scheidegger ◽  
G. Mariani ◽  
D. Nikolopoulos ◽  
C. Bekas ◽  
...  
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Large Scale ◽  
State Of The Art ◽  
Classification Performance ◽  
Neural Network Architecture ◽  
Network Information ◽  
Topological Network ◽  
Computational Resources

In recent years an increasing number of researchers and practitioners have been suggesting algorithms for large-scale neural network architecture search: genetic algorithms, reinforcement learning, learning curve extrapolation, and accuracy predictors. None of them, however, demonstrated highperformance without training new experiments in the presence of unseen datasets. We propose a new deep neural network accuracy predictor, that estimates in fractions of a second classification performance for unseen input datasets, without training. In contrast to previously proposed approaches, our prediction is not only calibrated on the topological network information, but also on the characterization of the dataset-difficulty which allows us to re-tune the prediction without any training. Our predictor achieves a performance which exceeds 100 networks per second on a single GPU, thus creating the opportunity to perform large-scale architecture search within a few minutes. We present results of two searches performed in 400 seconds on a single GPU. Our best discovered networks reach 93.67% accuracy for CIFAR-10 and 81.01% for CIFAR-100, verified by training. These networks are performance competitive with other automatically discovered state-of-the-art networks however we only needed a small fraction of the time to solution and computational resources.

Sign in / Sign up

Export Citation Format

Share Document