Artificial neural networks in structural dynamics: A new modular radial basis function approach vs. convolutional and feedforward topologies

Stoffel, Marcus; Gulakala, Rutwik; Bamer, Franz; Markert, Bernd

doi:10.1016/j.cma.2020.112989

Cited by 65 publications

(21 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, there is no universal design for the ANN that is applicable for any (or even a certain) type of experimental data [49]. Consequently, approaching the optimum design for the ANN is mainly a trial and error procedure [45] and also on the examples and experiences provided by the user [53]. Thus, the optimum design for the ANN can be proposed by varying the number of hidden layers, transfer function and the number of neurons in each hidden layer [17,56].…”

Section: Assigning the Functions And The Parameters Of The Annmentioning

confidence: 99%

See 1 more Smart Citation

Simulating the response of ionization chamber system to 137Cs irradiator using the artificial neural network modeling algorithm

Sadek

2020

SN Appl. Sci.

View full text Add to dashboard Cite

The artificial neural network algorithm has been used to simulate the response of an ionization chamber dosimetric system to the 137 Cs irradiator facility at the National Institute for Standards, Egypt. The performance of the designed model has been investigated over different processes functions and a wide range of parameters till the optimum performance has been approached. The standard uncertainty of the designed model has been evaluated for different dose-rate levels. It has been found that the dose-rate can be evaluated using the designed model with an uncertainty < 2.0% for the dose-rate level < 10 2 mGy/h and with an uncertainty < 0.5% for the dose-rate level [10 2 −10 3 ] mGy/h . The bias of the designed model has been evaluated and compared with the usual interpolation algorithms. The errors in the evaluated dose-rate using these methods have exceeded the 100% error level and approached the 20% error level for the dose rates less and greater than 10 mGy/h , respectively. While, the error in the evaluated dose-rate using the designed model did not exceed 7% and 2% for the same dose-rate levels, respectively.

show abstract

Section: Assigning the Functions And The Parameters Of The Annmentioning

confidence: 99%

“…On the other hand, the artificial neural network (ANN) is an intelligence method that is based on establishing a connection between the input and the output data for nonlinear system without the necessity to know how the system works internally [26]. Therefore, the ANN can lead to much lower computational time [53].…”

Section: Introductionmentioning

confidence: 99%

Simulating the response of ionization chamber system to 137Cs irradiator using the artificial neural network modeling algorithm

Sadek

2020

SN Appl. Sci.

View full text Add to dashboard Cite

show abstract

“…Lastly, a softmax layer containing 3 neurons was applied to classify the cells into three different classes: BMSCs (Control,0), chondrocytes (Negative,1), and treated group which are tenocytes differentiated from BMSCs (Positive,2). The rectified linear unit (ReLu) was applied as a nonlinear activation function to improve the performance of a CNNs [5], as well as it is the default activation for CNNs. The function will output the input directly if it is positive (2), otherwise, it will output zero (1).…”

Section: Convolutional Neural Networkmentioning

confidence: 99%

“…Deep neural networks, particularly the convolutional neural networks (CNNs), are widely applied in challenging imagebased classification by extracting image features from image pixels [5]. Many applications of deep neural networks have been reported in the field of cell biology including classification of myogenic C2C12 cells at differentiation [4] and red blood cells in sickle cell anemia [6], prediction of osteogenic differentiation potential [3], classification of intracellular actin networks [7], and evaluation of human-induced pluripotent stem cell [8].…”

Section: Introductionmentioning

confidence: 99%

Recognition of tenogenic differentiation using convolutional neural network

Dursun

Eschweiler

Tohidnezhad

et al. 2020

Current Directions in Biomedical Engineering

Self Cite

View full text Add to dashboard Cite

Methodologies to assess stem cell differentiation in the culturing state are needed for regenerative medicine and tissue engineering techniques. In recent years, convolutional neural networks (CNNs), a class of deep neural networks, have made impressive advancements in image-based classification, recognition and detection tasks. CNNs have been introduced as a non-invasive cell characterization method by learning features directly from image data of unlabeled cells. Furthermore, this approach serves as a rapid and inexpensive methodology with high performance compared to traditional techniques that require complex laboratory procedures including antibody staining and gene expression analysis. Here, we studied the potential of the CNNs approach to recognize stem cell differentiation based on cell morphology utilizing phasecontrast microscopy images.We have examined the differentiation potential of bone marrow mesenchymal stem cells (BMSCs) into tenocytes, with the treatment of bone morphogenetic protein-12 (BMP-12). After treatment, the phase-contrast images of cells were obtained directly from cell culture flasks to train CNN and the differentiated phenotype of stem cells was characterized by immunostaining. CNN was able to classify the cells into three groups including non-stem cells (chondrocytes), stem cells (BMSCs) and differentiated stem cells (tenocytes) based on their morphology with 92.2 % accuracy. The presented study revealed that CNN performed faster and non-invasive cell classification task compared to traditional methodologies.

show abstract

“…21 Based on certain input quantities, they have been used as surrogate models of nonlinear structures providing the numerical output values of interest. [22][23][24][25][26] More elaborate recurrent neural networks have shown to be useful for the description of time historydependent nonlinear behavior 27 as well as anisotropic elastoplasticity. 28 The incorporation of machine learning for structural design has been the focus of research for decades.…”

Section: Introductionmentioning

confidence: 99%

Machine‐learning‐enhanced tail end prediction of structural response statistics in earthquake engineering

Thaler

Stoffel

Markert

et al. 2021

Earthq Engng Struct Dyn

Self Cite

View full text Add to dashboard Cite

Evaluating the response statistics of nonlinear structures constitutes a key issue in engineering design. Hereby, the Monte Carlo method has proven useful, although the computational cost turns out to be considerably high. In particular, around the design point of the system near structural failure, a reliable estimation of the statistics is unfeasible for complex high-dimensional systems. Thus, in this paper, we develop a machine-learning-enhanced Monte Carlo simulation strategy for nonlinear behaving engineering structures. A neural network learns the response behavior of the structure subjected to an initial nonstationary ground excitation subset, which is generated based on the spectral properties of a chosen ground acceleration record. Then using the superior computational efficiency of the neural network, it is possible to predict the response statistics of the full sample set, which is considerably larger than the initial training sample set. To K E Y W O R D Searthquake generation, elastoplastic structures, Kanai-Tajimi filter, machine learning, Monte Carlo method, neural networks This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

show abstract

Artificial neural networks in structural dynamics: A new modular radial basis function approach vs. convolutional and feedforward topologies

Cited by 65 publications

References 36 publications

Simulating the response of ionization chamber system to 137Cs irradiator using the artificial neural network modeling algorithm

Simulating the response of ionization chamber system to 137Cs irradiator using the artificial neural network modeling algorithm

Recognition of tenogenic differentiation using convolutional neural network

Machine‐learning‐enhanced tail end prediction of structural response statistics in earthquake engineering

Contact Info

Product

Resources

About