Computational Statistics in Data Science. Группа авторов. Читать онлайн. Newlib. NEWLIB.NET

Автор: Группа авторов
Издательство: John Wiley & Sons Limited
Серия:
Жанр произведения: Математика
Год издания: 0
isbn: 9781119561088
Скачать книгу
v Subscript m Baseline right-parenthesis"/>, bold-italic upper W equals left-parenthesis bold-italic w 0 comma period period period comma bold-italic w Subscript m Baseline right-parenthesis, bold-italic x Subscript i Baseline equals left-parenthesis x Subscript i Superscript 0 Baseline comma x Subscript i Superscript 1 Baseline comma period period period comma x Subscript i Superscript p Baseline right-parenthesis, and tau left-parenthesis dot right-parenthesis and gamma left-parenthesis dot right-parenthesis are nonlinear functions.

stat08316fgz001

      3.3 Training an MLP

      We want to choose the weights and biases in such a way that they minimize the sum of squared errors within a given dataset. Similar to the general supervised learning approach, we want to find an optimal prediction delta Superscript asterisk Baseline left-parenthesis bold-italic upper X comma bold-italic upper W comma bold-italic upper V right-parenthesis such that

      where bold-italic upper X equals left-parenthesis bold-italic x 1 comma bold-italic x 2 comma period period period comma bold-italic x Subscript n Baseline right-parenthesis, and script l left-parenthesis dot comma dot right-parenthesis is cross‐entropy loss.

      (2)StartLayout 1st Row 1st Column script l left-parenthesis modifying above y with caret Subscript i Baseline comma y Subscript i Baseline right-parenthesis equals minus sigma-summation Underscript c equals 1 Overscript m Endscripts y Subscript i comma c Baseline log modifying above y with caret Subscript i comma c Baseline 2nd Column Blank EndLayout

      where m is the total number of classes; y Subscript i comma c Baseline equals 1 if the ith sample belongs to class c, otherwise it is equal to 0; and modifying above y with caret Subscript i comma c is the predicted score of the ith sample belonging to class c.

      We would like to address the issue of possibly being trapped in local minima, as backpropagation is a direct application of gradient descent to neural networks, and gradient descent is prone to finding local minima, especially in high‐dimensional spaces. It has been observed in practice that backpropagation actually does not typically get stuck in local minima and generally reaches the global minimum. There do, however, exist pathological data examples in which backpropagation will not converge to the global minimum, so convergence to the global minimum is certainly not an absolute guarantee. It remains a theoretical mystery why backpropagation does in fact generally converge to the global minimum, and under what conditions it will do so. However, some theoretical results have been developed to address this question. In particular, Gori and Tesi [7] established that for linearly separable data, backpropagation will always converge to the global solution.

      4.1 Introduction

      A CNN is a modified DNN that is particularly well equipped to handling image data. CNN usually contains not only fully connected layers but also convolutional layers and pooling layers, which make a difference. Image is a matrix of pixel values, which should be flattened to vectors before feeding into DNN as DNN takes a vector as input. However, spatial information might be lost in this process. The convolutional layer can take a matrix or tensor as input and is able to capture the spatial and temporal dependencies in an image.

      In the convolutional layer, the weight matrix (kernel) scans over the input image to produce a feature matrix. This process is called convolution operation. The pooling layer operates similar to the convolutional layer and