Features
Features:

Product Tour >

Edraw AI >

Paid Plans:

Individuals >

Business >

Eduaction >
Resources
Blog

History

How-tos & Tips

Discovery

Biography

Business Analysis

Examples

AI concept Map

Free AI Mind Map Generator

Onenote Mind Map

Bcg Matrix Examples

Nike Marketing Strategy

Unilever SWOT Analysis

Make Mind Maps in Google Docs

Guide

FAQs

What's New

Resource Center
Templates
All Templates

Brain Storming Templates

Strategy and Planning Templates

Project Management Templates

Product Management Templates

Human Resources Templates

Agile Workflow Templates

Marketing Templates

Education Templates

Fun and Games Templates

User Gallery
Download
Pricing
Enterprise

MindMap Gallery Handwritten digit recognition based on improved Sigmoid convolutional neural network

Handwritten digit recognition based on improved Sigmoid convolutional neural network

Since the amount of calculation is too large, the next_batch method is called to first filter a part of all training data sets, randomly select a part of the training data set, provide it to the input layer of the neural network, and then optimize the neural network through the reverse iteration method network

Edited at 2023-01-09 20:57:44

PlotWizard

Recent works View more works>>

Handwritten digit recognition based on improved Sigmoid convolutional neural network

PlotWizard

Recent works View more works>>

Recommended to you
Outline

Detailed explanation of the operation process of neural network
- 35
PlotWizard
Convolutional Neural Network (CNN)
- 104
PlotWizard
Common types and applications of neural network models
- 33
- 1
PlotWizard
Neural Networks and Deep Learning Recurrent Neural Networks
- 61
PlotWizard
Neural Networks and Deep Learning Convolutional Neural Networks
- 46
PlotWizard
The foundation of neural networks and deep learning
- 112
PlotWizard
Convolutional neural networks with dynamic regularization
- 64
PlotWizard
traditional neural network
- 20
PlotWizard
Neural Networks
- 1.0k
- 1
- 1
PlotWizard
Generate model
- 20
PlotWizard

Handwritten digit recognition based on improved Sigmoid convolutional neural network (2021 Information Technology)

chap1 model building

1) Mathematical neuron structure

xj is the input signal, f is the transfer function, wi,j represents the weight connected to neuron xj, yi represents the output value, and θ represents the threshold

2) BP neural network structure

In each round, the training results and the expected results are analyzed for errors, and then the weights and thresholds are modified, and a model whose output is consistent with the expected results is obtained step by step.

3) Model training process: When predicting images, Softmax Regression estimates a probability for each category and outputs the number with the highest probability as the result.

Find the optimal number of convolution kernels by testing small samples

The first layer uses the RELU function for activation; the second layer uses the Softmax function as the classifier to output the probability of each label.

Algorithm improvement based on CNN KNN

Fx is a trainable filter. Use fx to convolve an input image, and add a bias bx through the Sigmoid activation function to obtain the convolution layer Cx. Mj is the value of the input feature map

1) Gradient descent algorithm

2) Choose cross entropy as the loss function: the smaller the cross entropy, the better

The closer the model’s output is to the correct result

3) Algorithm process

① Define the algorithm formula, that is, the calculation during forward propagation of the neural network

② Specify the optimizer to optimize loss

③ Iteratively train the data

1) Calculate output

2) Calculate the loss function

3) Calculate the gradient through the loss function

4) Update weights through gradients

④ Import the training results into the KNN classifier and

The best nearest neighbor is obtained experimentally

⑤ Evaluate the accuracy on the test set or verification set

Comparison and analysis of experimental results

The impact of different values of neighbor number K on the accuracy of the algorithm, K=5 is the best

Compare with the recognition rates of several popular handwritten digit recognition algorithms

The recognition rate of the handwritten digits recognition algorithm based on KNN CNN used in this article is 99.7%, while the recognition rate of the handwritten digits recognition method based on the improved convolutional neural network is 99.2%, and the handwritten digits recognition method based on improved VGG-16 and Naive Bayes has a recognition rate of 99.2%. The recognition rate of the recognition method is 99.36%.