Series On Deep Learning Neural Networks - Böcker

This is the first book that uses Artificial Neural Networks (ANN) to solve fractional order systems. As a powerful data modeling tool, information is processed through neurons in parallel manner to solve a specific problem. Knowledge is acquired through learning and stored with inter neuron connections strength which are expressed by numerical values called weights. These weights are used to complete output signal values for new testing input signal value.In this book, multi-layer ANN model will be used to handle fractional order differential equations (FDEs). The network is trained using a back-propagation unsupervised learning algorithm which is based on the gradient descent rule. The ANN approximate solution of FDEs may be expressed as a sum of two terms; the first part satisfies boundary or initial conditions, and the second term contains ANN output with network parameters (weights and biases).Next, single layer Functional Link Artificial Neural Network (FLANN) models will be included for solving the FDEs. In FLANN the hidden layer is replaced by a functional expansion block for enhancement of the input patterns using orthogonal polynomials such as Chebyshev, Legendre, Hermite, etc. The computations become efficient because the procedure does not need to have hidden layer. Thus, the numbers of network parameters are less than the traditional ANN model.Varieties of FDEs will be addressed to show the reliability and efffectiveness of ANN. Singular nonlinear fractional Lane-Emden type equations, fractional vibration problems viz. Bagley-Torvik equations, fractional electrical problems viz. RLC, RC, LC circuit problems, Duffing oscillator problems with fractional derivatives etc. will be handled using multi-layer ANN and single layer FLANN models.

This comprehensive compendium designs deep neural network models and systems for intelligent analysis of fundus imaging. In response to several blinding fundus diseases such as Retinopathy of Prematurity (ROP), Diabetic Retinopathy (DR) and Macular Edema (ME), different image acquisition devices and fundus image analysis tasks are elaborated.From the actual fundus disease analysis tasks, various deep neural network models and experimental results are constructed and analyzed. For each task, an actual system for clinical application is developed.This useful reference text provides theoretical and experimental reference basis for AI researchers, system engineers of intelligent medicine and ophthalmologists.