How does softmax classify the data of the fully connected layer?
tags: softmax Fully connected layer cnn Deep learning Neural Networks
Fully connected layer with softmax
There are many nodes in the fully connected layer. How does softmax classify the data of the fully connected layer? The following is personal ignorance.
When doing deep learning, such as CNN network, the end of the network is often the full connection layer and the softmax layer.
In the figure, W represents the weight matrix, bias is the offset, X is the input from the fully connected layer, and Z is the column vector of K. The figure below more intuitively describes the process of softmax classification.
Wj is regarded as the weight of the feature under the j-th class, that is, the importance degree of each dimension feature and the degree of influence on the final score, and then mapped to probability by Softmax. The specific operation process is as follows
It can be seen that there are three categories in the above figure, softmax maps the original input value to (0,1), the probability of belonging to the Z1 class is 0.88, and the probability of belonging to the Z2 class is 0.12. The probability of belonging to the Z3 class is approximately equal to 0, and the sum of all class probabilities is 1. Classification result takes its maximum value
Intelligent Recommendation
Understanding of the fully connected layer
Derivation of the fully connected layer Each node of the fully connected layer is connected to all nodes of the previous layer to combine the features extracted from the front side. Due to its fully c...
tf.keras.layers.Dense fully connected layer
units: Positive integer, dimensionality of the output space. activation: Activation function to use. If you don’t specify anything, no activation is applied (ie. “linear” activation:...
The role of the fully connected layer
Turn to know Author: Wei show participation link: https://www.zhihu.com/question/41037974/answer/150522307 What is the purpose of the fully connected layer? Let me talk about three points. Fully conne...
Fully connected layer detailed
Note: This series of blogs is to summarize the essence of CSDN posts, similar to self-use notes, no learning exchanges, to facilitate future review and review, the citations in the blog posts are all ...
Tensorflow2.0 fully connected layer
Basic use of fully connected layer Use functions to create a fully connected network...
More Recommendation
Understanding of fully connected layers, softmax and softmax loss in convolutional neural networks
Generally, a CNN network mainly includes a convolution layer, a pooling layer, a full connection layer, a loss layer, and the like. The convolutional layer and the pooling layer are not understood in ...
Softmax method to classify CIFAR-10 data
Softmax method to classify CIFAR-10 data Codeļ¼ I don't know why the comment part is always wrong, just copy the loss and dw calculation code of others directly later. I will check it later svm is very...
Data loading of tensorflow 2.0 neural network and fully connected layer
5.1 Data loading Outline keras.dataset MNIST data set CIFAR10/100 tf.data.Dataset from_tensor_slices() .shuff .map .batch StopIteration .repeat Outline keras.datasets tf.data.Dataset.from_tensor_slice...
Tensorflow notes-data set loading, fully connected layer, output method
One. Data Loading (DataLoading) Kreas.datasets: Commonly used small data sets provided by tensorflow; Tf.data.Dataset.from_tensor_slices: Load the data into tensor type into memory; two. Full Connecti...
Convolutional layer, pooling layer and fully connected layer
Convolution layer (Convolution) What is convolution, as shown in the figure below, the convolution operation uses a convolution kernel to move regularly on a picture according to a certain step size. ...