scholarly journals Reconstruction of 3D Object Shape Using Hybrid Modular Neural Network Architecture Trained on 3D Models from ShapeNetCore Dataset

Sensors ◽  
2019 ◽  
Vol 19 (7) ◽  
pp. 1553 ◽  
Author(s):  
Audrius Kulikajevas ◽  
Rytis Maskeliūnas ◽  
Robertas Damaševičius ◽  
Sanjay Misra
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
A Priori ◽  
Three Dimensional ◽  
Qualitative Evaluation ◽  
3D Models ◽  
Depth Sensors ◽  
The Neural Network ◽  
Artificial Neural Network Ann ◽  
Commercial Applications

Depth-based reconstruction of three-dimensional (3D) shape of objects is one of core problems in computer vision with a lot of commercial applications. However, the 3D scanning for point cloud-based video streaming is expensive and is generally unattainable to an average user due to required setup of multiple depth sensors. We propose a novel hybrid modular artificial neural network (ANN) architecture, which can reconstruct smooth polygonal meshes from a single depth frame, using a priori knowledge. The architecture of neural network consists of separate nodes for recognition of object type and reconstruction thus allowing for easy retraining and extension for new object types. We performed recognition of nine real-world objects using the neural network trained on the ShapeNetCore model dataset. The results evaluated quantitatively using the Intersection-over-Union (IoU), Completeness, Correctness and Quality metrics, and qualitative evaluation by visual inspection demonstrate the robustness of the proposed architecture with respect to different viewing angles and illumination conditions.

Stable and accurate a posteriori LES of 2D turbulence with convolutional neural networks: Backscatter analysis and generalization via transfer learning

2021 ◽  
Author(s):  
Yifei Guan ◽  
Ashesh Chattopadhyay ◽  
Adam Subel ◽  
Pedram Hassanzadeh
Keyword(s):  
Neural Network ◽  
Transfer Learning ◽  
Network Architecture ◽  
Turbulence Modeling ◽  
Scale Effects ◽  
A Priori ◽  
A Posteriori ◽  
Subgrid Scale ◽  
Artificial Neural Network Ann ◽  
Short Term Prediction

<p>In large eddy simulations (LES), the subgrid-scale effects are modeled by physics-based or data-driven methods. This work develops a convolutional neural network (CNN) to model the subgrid-scale effects of a two-dimensional turbulent flow. The model is able to capture both the inter-scale forward energy transfer and backscatter in both a priori and a posteriori analyses. The LES-CNN model outperforms the physics-based eddy-viscosity models and the previous proposed local artificial neural network (ANN) models in both short-term prediction and long-term statistics. Transfer learning is implemented to generalize the method for turbulence modeling at higher Reynolds numbers. Encoder-decoder network architecture is proposed to generalize the model to a higher computational grid resolution.</p>

scholarly journals An Artificial Neural Network Based Solution Scheme for Periodic Computational Homogenization of Electrostatic Problems

2019 ◽  
Vol 24 (2) ◽  
pp. 40 ◽  
Author(s):  
Felix Selim Göküzüm ◽  
Lu Trong Khiem Nguyen ◽  
Marc-André Keip
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Trial Function ◽  
A Priori ◽  
Three Dimensional ◽  
Internal Parameters ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Core Idea ◽  
Electrostatic Problems

The present work addresses a solution algorithm for homogenization problems based on an artificial neural network (ANN) discretization. The core idea is the construction of trial functions through ANNs that fulfill a priori the periodic boundary conditions of the microscopic problem. A global potential serves as an objective function, which by construction of the trial function can be optimized without constraints. The aim of the new approach is to reduce the number of unknowns as ANNs are able to fit complicated functions with a relatively small number of internal parameters. We investigate the viability of the scheme on the basis of one-, two- and three-dimensional microstructure problems. Further, global and piecewise-defined approaches for constructing the trial function are discussed and compared to finite element (FE) and fast Fourier transform (FFT) based simulations.

Constrained Hybrid Neural Modelling of Biotechnological Processes

2010 ◽  
Vol 8 (1) ◽  
Author(s):  
Asma Karama ◽  
Olivier Bernard ◽  
Jean-Luc Gouzé
Keyword(s):  
Neural Network ◽  
Mass Balance ◽  
Network Architecture ◽  
A Priori ◽  
Neural Model ◽  
Neural Network Architecture ◽  
The Neural Network ◽  
Wide Range ◽  
Neural Modelling ◽  
Biotechnological Processes

We propose a general methodology to develop a hybrid neural model for a wide range of biotechnological processes. The hybrid neural modelling approach combines the flexibility of a neural network representation of unknown process kinetics with a global mass-balance based process description. The hybrid model is built in such a way that its trajectories keep their physical and biological meaning (mass balance, positivity of the concentrations, boundness, saturation or inhibition of kinetics) even far from the identification data conditions. We examine the constraints (a priori knowledge) that must be satisfied by the model and that provide additional conditions to be imposed on the neural network. We illustrate our approach with various biotechnological processes showing how to select the appropriate neural network architecture. The method is detailed for modelling an anaerobic wastewater treatment bioreactor using experimental data.

SCORING MODELING BASED ON NEURAL NETWORKS FOR DETERMINING A BANK BORROWER'S RATING

2020 ◽  
Vol 2020 (10) ◽  
pp. 54-62
Author(s):  
Oleksii VASYLIEV ◽  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Network Architecture ◽  
Statistical Data ◽  
Activation Function ◽  
Decision Making Process ◽  
Neural Network Architecture ◽  
Acceptable Accuracy ◽  
The Neural Network ◽  
Sigmoid Activation Function

The problem of applying neural networks to calculate ratings used in banking in the decision-making process on granting or not granting loans to borrowers is considered. The task is to determine the rating function of the borrower based on a set of statistical data on the effectiveness of loans provided by the bank. When constructing a regression model to calculate the rating function, it is necessary to know its general form. If so, the task is to calculate the parameters that are included in the expression for the rating function. In contrast to this approach, in the case of using neural networks, there is no need to specify the general form for the rating function. Instead, certain neural network architecture is chosen and parameters are calculated for it on the basis of statistical data. Importantly, the same neural network architecture can be used to process different sets of statistical data. The disadvantages of using neural networks include the need to calculate a large number of parameters. There is also no universal algorithm that would determine the optimal neural network architecture. As an example of the use of neural networks to determine the borrower's rating, a model system is considered, in which the borrower's rating is determined by a known non-analytical rating function. A neural network with two inner layers, which contain, respectively, three and two neurons and have a sigmoid activation function, is used for modeling. It is shown that the use of the neural network allows restoring the borrower's rating function with quite acceptable accuracy.

Spatial Variability Aware Deep Neural Networks (SVANN): A General Approach

2021 ◽  
Vol 12 (6) ◽  
pp. 1-21
Author(s):  
Jayant Gupta ◽  
Carl Molnar ◽  
Yiqun Xie ◽  
Joe Knight ◽  
Shashi Shekhar
Keyword(s):  
Neural Network ◽  
Spatial Variability ◽  
Network Architecture ◽  
Network Models ◽  
Neural Network Architecture ◽  
Neural Network Models ◽  
Climatic Zones ◽  
The Neural Network ◽  
Plant Hardiness ◽  
Interpretation Model

Spatial variability is a prominent feature of various geographic phenomena such as climatic zones, USDA plant hardiness zones, and terrestrial habitat types (e.g., forest, grasslands, wetlands, and deserts). However, current deep learning methods follow a spatial-one-size-fits-all (OSFA) approach to train single deep neural network models that do not account for spatial variability. Quantification of spatial variability can be challenging due to the influence of many geophysical factors. In preliminary work, we proposed a spatial variability aware neural network (SVANN-I, formerly called SVANN ) approach where weights are a function of location but the neural network architecture is location independent. In this work, we explore a more flexible SVANN-E approach where neural network architecture varies across geographic locations. In addition, we provide a taxonomy of SVANN types and a physics inspired interpretation model. Experiments with aerial imagery based wetland mapping show that SVANN-I outperforms OSFA and SVANN-E performs the best of all.

An artificial neural network approach to discrete-event simulation

1995 ◽  
Vol 9 (1) ◽  
pp. 37-49 ◽  
Author(s):  
Ian Flood ◽  
Kenneth Worley
Keyword(s):  
Neural Network ◽  
Discrete Event Simulation ◽  
Network Architecture ◽  
Discrete Event ◽  
Network Approach ◽  
Neural Network Approach ◽  
Event Simulation ◽  
The Neural Network ◽  
Component Processes ◽  
And Performance

AbstractThis paper proposes and evaluates a neural network-based method for simulating manufacturing processes that exhibit both noncontinuous and stochastic behavior processes more conventionally modeled, using discrete-event simulation algorithms. The incentive for developing the technique is its potential for rapid execution of a simulation through parallel processing, and facilitation of the development and improvement of models particularly where there is limited theory describing the dependence between component processes. A brief introduction is provided to a radial-Gaussian neural network architecture and training process, the system adopted for the work presented in this paper. A description of the basic approach proposed for applying this technology to simulation is then described. This involves the use of a modularized neural network approach to model construction and the prediction of the occurrence of events using information retained from several previous states of the simulation. A class of earth-moving systems, comprising a push-dozer and a fleet of scrapers, is used as the basis for assessing the viability and performance of the proposed approach. A series of experiments show the neural network to be capable of both capturing the characteristic behavior and making an accurate prediction of production rates of scraper-based earth-moving systems. The paper concludes with an indication of some areas for further development and evaluation of the technique.

A Model on the Correlation between Composition and Mechanical Properties of Mg-Al-Zn Alloys by Using Artificial Neural Network

2005 ◽  
Vol 488-489 ◽  
pp. 793-796 ◽  
Author(s):  
Hai Ding Liu ◽  
Ai Tao Tang ◽  
Fu Sheng Pan ◽  
Ru Lin Zuo ◽  
Ling Yun Wang
Keyword(s):  
Neural Network ◽  
Mechanical Properties ◽  
Artificial Neural Network ◽  
Magnesium Alloys ◽  
Tensile Yield Strength ◽  
Data Set ◽  
The Neural Network ◽  
Artificial Neural ◽  
Prediction Of Mechanical Properties ◽  
Artificial Neural Network Ann

A model was developed for the analysis and prediction of correlation between composition and mechanical properties of Mg-Al-Zn (AZ) magnesium alloys by applying artificial neural network (ANN). The input parameters of the neural network (NN) are alloy composition. The outputs of the NN model are important mechanical properties, including ultimate tensile strength, tensile yield strength and elongation. The model is based on multilayer feedforward neural network. The NN was trained with comprehensive data set collected from domestic and foreign literature. A very good performance of the neural network was achieved. The model can be used for the simulation and prediction of mechanical properties of AZ system magnesium alloys as functions of composition.

scholarly journals ON SOME BOUNDS OF THE MINIMUM EDGE DOMINATING ENERGY OF A GRAPH

2021 ◽  
Vol 12 (4) ◽  
pp. 1285-1296
Author(s):  
Dr. Gauri Ghule , Et. al.
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Fuzzy Controller ◽  
Back Propagation ◽  
Back Propagation Neural Network ◽  
Superior Performance ◽  
Fuzzy Classifier ◽  
Manual Intervention ◽  
The Neural Network ◽  
Hidden Neurons

Number of hidden neurons is necessary constant for tuning the neural network to achieve superior performance. These parameters are set manually through experimentation. The performance of the network is evaluated repeatedly to choose the best input parameters.Random selection of hidden neurons may cause underfitting or overfitting of the network. We propose a novel fuzzy controller for finding the optimal value of hidden neurons automatically. The hybrid classifier helps to design competent neural network architecture, eliminating manual intervention for setting the input parameters. The effectiveness of tuning the number of hidden neurons automatically on the convergence of a back-propagation neural network, is verified on speech data. The experimental outcomes demonstrate that the proposed Neuro-Fuzzy classifier can be viably utilized for speech recognition with maximum classification accuracy.

scholarly journals A SELF-ADAPTABLE DISTRIBUTED CBR VERSION OF THE EQUIVOX SYSTEM

2016 ◽  
Vol 28 (04) ◽  
pp. 1650028
Author(s):  
Julien Henriet ◽  
Christophe Lang ◽  
Ronnie Muthada Pottayya ◽  
Karla Breschi
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Three Dimensional ◽  
Multi Agent System ◽  
Accidental Exposure ◽  
Interpolation Accuracy ◽  
Multi Agent ◽  
Artificial Neural Network Ann ◽  
Numerical Representations ◽  
Voxel Phantoms

Three dimensional (3D) voxel phantoms are numerical representations of human bodies, used by physicians in very different contexts. In the controlled context of hospitals, where from 2 to 10 subjects may arrive per day, phantoms are used to verify computations before therapeutic exposure to radiation of cancerous tumors. In addition, 3D phantoms are used to diagnose the gravity of accidental exposure to radiation. In such cases, there may be from 10 to more than 1000 subjects to be diagnosed simultaneously. In all of these cases, computation accuracy depends on a single such representation. In this paper, we present EquiVox which is a tool composed of several distributed functions and enables to create, as quickly and as accurately as possible, 3D numerical phantoms that fit anyone, whatever the context. It is based on a multi-agent system. Agents are convenient for this kind of structure, they can interact together and they may have individual capacities. In EquiVox, the phantoms adaptation is a key phase based on artificial neural network (ANN) interpolations. Thus, ANNs must be trained regularly in order to take into account newly capitalized subjects and to increase interpolation accuracy. However, ANN training is a time-consuming process. Consequently, we have built Equivox to optimize this process. Thus, in this paper, we present our architecture, based on agents and ANN, and we put the stress on the adaptation module. We propose, next, some experimentations in order to show the efficiency of the EquiVox architecture.

scholarly journals Neural Networks Applied to Estimating Subglacial Topography and Glacier Volume

2009 ◽  
Vol 22 (8) ◽  
pp. 2146-2160 ◽  
Author(s):  
Garry K. C. Clarke ◽  
Etienne Berthier ◽  
Christian G. Schoof ◽  
Alexander H. Jarosch
Keyword(s):  
Neural Network ◽  
Regional Scale ◽  
Ice Thickness ◽  
Ice Dynamics ◽  
Ice Caps ◽  
Mountain Glaciers ◽  
The Neural Network ◽  
Glacier Ice ◽  
Subglacial Topography ◽  
Artificial Neural Network Ann

Abstract To predict the rate and consequences of shrinkage of the earth’s mountain glaciers and ice caps, it is necessary to have improved regional-scale models of mountain glaciation and better knowledge of the subglacial topography upon which these models must operate. The problem of estimating glacier ice thickness is addressed by developing an artificial neural network (ANN) approach that uses calculations performed on a digital elevation model (DEM) and on a mask of the present-day ice cover. Because suitable data from real glaciers are lacking, the ANN is trained by substituting the known topography of ice-denuded regions adjacent to the ice-covered regions of interest, and this known topography is hidden by imagining it to be ice-covered. For this training it is assumed that the topography is flooded to various levels by horizontal lake-like glaciers. The validity of this assumption and the estimation skill of the trained ANN is tested by predicting ice thickness for four 50 km × 50 km regions that are currently ice free but that have been partially glaciated using a numerical ice dynamics model. In this manner, predictions of ice thickness based on the neural network can be compared to the modeled ice thickness and the performance of the neural network can be evaluated and improved. From the results, thus far, it is found that ANN depth estimates can yield plausible subglacial topography with a representative rms elevation error of ±70 m and remarkably good estimates of ice volume.

Sign in / Sign up

Export Citation Format

Share Document