For more information: asmedigitalcol...
Machine learning and deep learning approaches are finally catching up with Material science community. Molecular simulations and property predictions can be made with deep neural networks.However, conventional neural networks such as perceptron, convoluted neural networks (CNN) and recurrent neural networks (RNN) do not provide accurate solution to molecular and material problems. Novel special purpose neural network architectures need to be designed for molecules and materials. In this video, primary requirements and assumptions are discussed that can help the users design deep model for their molecular systems. Further primary concepts of high dimensional neural networks (BPNN), that are extensively used for energy prediction of molecular systems, has been defined along with its limitations. BPNN based approach is currently used in many recently developed models.
To go through this video, view must have prior knowledge of basics of artificial neural networks and their training process.
Some useful links: • But what is a neural n...
More about BPNN neural network, force/atom calculations can be found at:
aip.scitation....
contact: bingeatoms@gmail.com
Негізгі бет Introduction to Neural Network Architectures for Molecular Systems
Пікірлер: 6