http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A neural network approach for simulating stationary stochastic processes
Beer, Michael,Spanos, Pol D. Techno-Press 2009 Structural Engineering and Mechanics, An Int'l Jou Vol.32 No.1
In this paper a procedure for Monte Carlo simulation of univariate stationary stochastic processes with the aid of neural networks is presented. Neural networks operate model-free and, thus, circumvent the need of specifying a priori statistical properties of the process, as needed traditionally. This is particularly advantageous when only limited data are available. A neural network can capture the "pattern" of a short observed time series. Afterwards, it can directly generate stochastic process realizations which capture the properties of the underlying data. In the present study a simple feed-forward network with focused time-memory is utilized. The proposed procedure is demonstrated by examples of Monte Carlo simulation, by synthesis of future values of an initially short single process record.
A neural network approach for simulating stationary stochastic processes
Michael Beer,Pol D. Spanos 국제구조공학회 2009 Structural Engineering and Mechanics, An Int'l Jou Vol.32 No.1
In this paper a procedure for Monte Carlo simulation of univariate stationary stochastic processes with the aid of neural networks is presented. Neural networks operate model-free and, thus, circumvent the need of specifying a priori statistical properties of the process, as needed traditionally. This is particularly advantageous when only limited data are available. A neural network can capture the “pattern” of a short observed time series. Afterwards, it can directly generate stochastic process realizations which capture the properties of the underlying data. In the present study a simple feedforward network with focused time-memory is utilized. The proposed procedure is demonstrated by examples of Monte Carlo simulation, by synthesis of future values of an initially short single process record.
Seismic Response Meta-model of High-Rise Fame Structure Based on Time-Delay Neural Network
He Zhang,Marius Bittner,Michael Beer 대한토목학회 2022 KSCE JOURNAL OF CIVIL ENGINEERING Vol.26 No.5
To make structural seismic response simulation more efficient, a meta-model method which is based on the time delay neural network is proposed. And an accuracy evaluation method that considers the drift peak amplitudes and maximum amplitudes in each intensity as performance parameters is also proposed, this method can make a balance between accuracy and training time. Exampled by 4 frame structures which are all 20 stories, and accuracy evaluating results show that more than 80% of samples, which include training models and testing models of these performance parameters can be explained by meta models’ fitting. The average time to simulate by this method is 0.08s and faster than the finite element method which spends 24 min averagely.
Right Hemispheric Predominance of Brain Infarcts in Atrial Fibrillation: A Lesion Mapping Analysis
Anna Altermatt,Tim Sinnecker,Stefanie Aeschbacher,Anne Springer,Michael Coslovsky,Juerg Beer,Giorgio Moschovitis,Angelo Auricchio,Urs Fischer,Carole E. Aubert,Michael Kühne,David Conen,Stefan Osswald 대한뇌졸중학회 2022 Journal of stroke Vol.24 No.1
.
Shairah Radzi,Heang Kuan Joel Tan,Gerald Jit Shen Tan,Wai Yee Yeong,Michael Alan Ferenczi,Naomi Low-Beer,Sreenivasulu Reddy Mogali 대한해부학회 2020 Anatomy & Cell Biology Vol.53 No.1
Learning anatomy is commonly facilitated by use of cadavers, plastic models and more recently three-dimensional printed (3DP) anatomical models as they allow students to physically touch and hold the body segments. However, most existing models are limited to surface features of the specimen, with little opportunity to manipulate the structures. There is much interest in developing better 3DP models suitable for anatomy education. This study aims to determine the feasibility of developing a multi-material 3DP heart model, and to evaluate students’ perceptions of the model. Semi-automated segmentation was performed on computed tomgoraphy plastinated heart images to develop its 3D digital heart model. Material jetting was used as part of the 3D printing process so that various colors and textures could be assigned to the individual segments of the model. Morphometric analysis was conducted to quantify the differences between the printed model and the plastinated heart. Medical students’ opinions were sought using a 5-point Likert scale. The 3DP full heart was anatomically accurate, pliable and compressible to touch. The major vessels of the heart were color-coded for easy recognition. Morphometric analysis of the printed model was comparable with the plastinated heart. Students were positive about the quality of the model and the majority of them reported that the model was useful for their learning and that they would recommend their use for anatomical education. The successful feasibility study and students’ positive views suggest that the development of multi-material 3DP models is promising for medical education.