The nonlinear activation functions in the deep CNN(Convolutional Neural Network)based on fluid dynamics are presented.We propose two types of activation functions by applying the so-called parametric softsign to the n...The nonlinear activation functions in the deep CNN(Convolutional Neural Network)based on fluid dynamics are presented.We propose two types of activation functions by applying the so-called parametric softsign to the negative region.We use significantly the well-known TensorFlow as the deep learning framework.The CNN architecture consists of three convolutional layers with the max-pooling and one fullyconnected softmax layer.The CNN approaches are applied to three benchmark datasets,namely,MNIST,CIFAR-10,and CIFAR-100.Numerical results demonstrate the workability and the validity of the present approach through comparison with other numerical performances.展开更多
With the increasing growth of AI,big data analytics,cloud computing,and Internet of Things applications,devel-oping memristor devices and related hardware systems to compute the deep learning application needs extensi...With the increasing growth of AI,big data analytics,cloud computing,and Internet of Things applications,devel-oping memristor devices and related hardware systems to compute the deep learning application needs extensive data calculations with low power consumption and lesser chip area.Deep learning model is one of the AI methods which is gaining importance in object detection,natural language processing,and pattern recognition.A large amount of data handling is essential for driving the deep learning model with less power consumption.To address these issues,the paper proposed the Memristor-based object detection on the CIFAR-10 dataset and achieved an accuracy of 85 percent.The memtorch package in python is employed to predict the objects for implementation.展开更多
文摘The nonlinear activation functions in the deep CNN(Convolutional Neural Network)based on fluid dynamics are presented.We propose two types of activation functions by applying the so-called parametric softsign to the negative region.We use significantly the well-known TensorFlow as the deep learning framework.The CNN architecture consists of three convolutional layers with the max-pooling and one fullyconnected softmax layer.The CNN approaches are applied to three benchmark datasets,namely,MNIST,CIFAR-10,and CIFAR-100.Numerical results demonstrate the workability and the validity of the present approach through comparison with other numerical performances.
文摘With the increasing growth of AI,big data analytics,cloud computing,and Internet of Things applications,devel-oping memristor devices and related hardware systems to compute the deep learning application needs extensive data calculations with low power consumption and lesser chip area.Deep learning model is one of the AI methods which is gaining importance in object detection,natural language processing,and pattern recognition.A large amount of data handling is essential for driving the deep learning model with less power consumption.To address these issues,the paper proposed the Memristor-based object detection on the CIFAR-10 dataset and achieved an accuracy of 85 percent.The memtorch package in python is employed to predict the objects for implementation.
